FDA Briefing Document

ARTHRITIS ADVISORY COMMITTEE AND DRUG SAFETY AND RISK MANAGEMENT ADVISORY COMMITTEE MEETING January 11, 2019

NDA 21856 Febuxostat Xanthine oxidase (XO) inhibitor for the chronic management of hyperuricemia in patients with gout

Takeda

1 of 169

DISCLAIMER STATEMENT

The attached package contains background information prepared by the Food and Drug Administration (FDA) for the panel members of the advisory committee. The FDA background package often contains assessments and/or conclusions and recommendations written by individual FDA reviewers. Such conclusions and recommendations do not necessarily represent the final position of the individual reviewers, nor do they necessarily represent the final position of the Review Division or Office. We bring the supplemental new drug application (sNDA) for Uloric (febuxostat) NDA# 21856 which includes the results from the post-marketing safety trial required by FDA to evaluate the cardiovascular (CV) safety of febuxostat - the Cardiovascular Safety of Febuxostat and Allopurinol in Patients with Gout and Cardiovascular Morbidities (CARES) trial to this Advisory Committee to gain the Committee’s insights and opinions. The background package may not include all issues relevant to the final regulatory recommendation and instead is intended to focus on issues identified by the Agency for discussion by the advisory committee. The FDA will not issue a final determination on the issues at hand until input from the advisory committee process has been considered and all reviews have been finalized. The final determination may be affected by issues not discussed at the advisory committee meeting.

2 of 169 1 Table of Contents

1 Division Memorandum ……………………………………………………………………………….... 6

2 Draft Points to Consider …………………………………………………………………………………. 17

3 Clinical and Statistical Review ………………………………………………………………………….. 18

4 Division of Cardiovascular and Renal Products Consult ……………………………………. 99

5 Office of Surveillance and Epidemiology Review …………………………………………….. 108

6 Office of Surveillance and Epidemiology – Sentinel Memo ……………………………… 138

3 of 169 Table of Tables Table 1 Primary Analysis of MACE – On Study Analysis...... 11 Table 2 Analysis of CV Death and All Cause Death in CARES ...... 12 Table 3: Treatments for the Management of Hyperuricemia ...... 22 Table 4 Summary of Major Safety Results from First Cycle Review of Febuxostat ...... 26 Table 5: All-Cause Mortality in Febuxostat Safety Database by Patient-Years of Exposure (Data Cut-off 08 February 2006) ...... 27 Table 6: Clinical Trials Relevant to this NDA ...... 31 Table 7 – Tabular Summary of Major Inclusion and Exclusion Criteria for Study 301 ...... 35 Table 8 - Tabular Summary of Assessments and Procedures for Study 301 ...... 38 Table 9: Cardiovascular Endpoint Definitions Used to Adjudicate Deaths Reported in Study 301 .... 40 Table 10. Subjects Follow-up Until Trial Discontinuation (FAS) ...... 48 Table 11. Exposure Time in Subjects Exposed to Study Treatment (FAS) ...... 49 Table 12. Disposition of Subjects by Vital Status ...... 50 Table 13 Follow-up Time and Treatment Exposure in Subjects Who Prematurely Discontinued Study Treatment ...... 50 Table 14 Follow-up Time and Treatment Exposure in Subjects Who Prematurely Discontinued Study Visits ...... 51 Table 15: Summary of Protocol Violations/Deviations for Study 301 ...... 51 Table 16: Baseline Demographics of Subjects Enrolled in Study 301 ...... 52 Table 17: Comorbid Disease Risk Factors for Subjects Enrolled in Study 301 ...... 53 Table 18: Summary of Subjects' Gout History, Disease Status and Treatment History for Study 301 ...... 54 Table 19: Summary of Subjects' Cardiovascular History in Study 301 ...... 55 Table 20: Subjects' Compliance with Randomized Study Medication for Study 301 ...... 56 Table 21: Summary of Study Drug Dose Adjustments for Study 301 (All Subjects) ...... 57 Table 22: Prophylactic Gout Medication Use and Dose Adjustments for Study 301 ...... 57 Table 23: Concomitant Medications Taken by > 10% of Subjects in Study 301 ...... 58 Table 24 Primary Analysis of MACE – Number of Events ...... 60 Table 25. Secondary Analysis of Cardiovascular Endpoints – Number of Events ...... 62 Table 26. Number Needed to Harm in Study TMX-301 ...... 63 Table 27. Number of Event by the Windows of Time after Treatment Discontinuation ...... 64 Table 28. Analysis of CV Death and All-Cause Death – Number of Events ...... 65 Table 29. Sensitivity Analysis by Renal Function Status and Final Dose of Treatment Drug ...... 66 Table 30. Additional 199 deaths by Windows of Time after Treatment Discontinuation...... 67 Table 31. Follow-up Time and Treatment Exposure Time for Additional 199 Deaths ...... 68 Table 32. Follow-up Time in 821 Subjects with Unknown Vital Status ...... 68 Table 33. Primary and Secondary Analyses of Cardiovascular Endpoints ...... 72 Table 34: Summary of Subjects by Treatment Group Who Experienced Treatment Emergent Adverse Events and Deaths During Study 301...... 74 Table 35: Treatment Emergent Adverse Events Leading to Death by MedDRA System Organ Class and Preferred Term by Treatment Group for Study 301 ...... 75

4 of 169 Table 36: Causes of Adjudicated Cardiovascular Deaths Reported in Study 301 by Various Time Windows ...... 79 Table 37: Summary of Adjudicated Non-Cardiovascular Deaths by MedDRA Preferred Term by Treatment Group for Study 301 ...... 80 Table 38: Abridged Summary of Treatment-Emergent Serious Adverse Events (SAEs) by MedDRA System Organ Class and Preferred Term Occurring in >1% of Subjects by Randomized Treatment Group for Study 301 ...... 81 Table 39: Treatment Emergent Adverse Events Leading to Discontinuation of Randomized Study Medication by MedDRA System Organ Class/Preferred Term Occurring in > 0.1% of Subjects by Randomized Treatment Group for Study 301 ...... 85 Table 40: Summary of Severe Treatment Emergent Adverse Events (TEAEs)1 by MedDRA System Organ Class in Study 3012 ...... 87 Table 41: Treatment Emergent Adverse Events1 (TEAEs) by MedDRA System Organ Class by Treatment Group for Study 3012 ...... 88 Table 42: Summary of Common Treatment Emergent Adverse Events1 by Preferred Term Occurring in >5% of Subjects by Treatment Group in Study 3012,3 ...... 89 Table 43: Proportion of Subjects with Serum Uric Acid Levels <6.0 mg/dL and <5.0 mg/dL by Visit by Treatment Group in Study 301 (Full Analysis Set) ...... 90 Table 44: Proportion of Subjects with Gout Flares Requiring Treatment During Study 301 (Full Analysis Set) ...... 91

Table of Figures Figure 1 Schema of Study 301 ...... 35 Figure 2.Disposition of Subjects in Trial TMX-301 ...... 47 Figure 3. Discontinuations from Study Visits ...... 48 Figure 4. Study Follow-up Until Trial Discontinuation ...... 49 Figure 5. Kaplan-Meier Cumulative Probability of Primary MACE by Treatment ...... 60 Figure 6. Kaplan-Meier Plot for Cardiovascular Death ...... 61 Figure 7. Kaplan-Meier Plot for All-Cause Death ...... 63 Figure 8. Forest Plot of Hazard Ratio for MACE by Baseline Characteristics ...... 69 Figure 9. Forest Plot of Hazard Ratio for CV Death by Baseline Subgroups (1) ...... 70 Figure 10. Forest Plot of Hazard Ratio for CV Death by Baseline Subgroups (2) ...... 71 Figure 11. Assessment of Proportional Hazards Assumption: Schoenfeld Residual Plot (MACE) 97 Figure 12. Assessment of Proportional Hazards Assumption: Schoenfeld Residual Plot (CV Death) ...... 97 Figure 13. Assessment of Proportional Hazards Assumption: Schoenfeld Residual Plot (All-Cause Death) ...... 98

5 of 169

2 Division Memorandum

Department of Health and Human Services Public Health Service Food and Drug Administration Center for Drug Evaluation and Research M E M O R A N D U M

Date: December 14, 2018

From: Sally Seymour, MD Acting Director, Division of Pulmonary, Allergy, and Rheumatology Products (DPARP)

Nikolay Nikolov, MD Associate Director for Rheumatology, DPARP

To: Arthritis Advisory Committee (AAC) and Drug Safety and Risk Management Advisory Committee (DSaRM)

Subject: Overview of the January 11, 2019, Joint AAC and DSaRM meeting

Thank you for your participation in the upcoming joint meeting of the Arthritis Advisory Committee (AAC) and Drug Safety and Risk Management (DSaRM) Advisory Committee, to be held January 11, 2019. As members of FDA Advisory Committees (AC), we consider your expert scientific advice and recommendations to the FDA very important to our regulatory decision- making processes. The objective of the upcoming meeting is to discuss the supplemental new drug application (sNDA) for Uloric (febuxostat), NDA# 21856, which provides for the results from the post-marketing safety trial required by FDA to evaluate the cardiovascular (CV) safety of febuxostat - the Cardiovascular Safety of Febuxostat and Allopurinol in Patients with Gout and Cardiovascular Morbidities (CARES) trial. The discussion will include the results from the CARES trial, the benefit risk assessment of febuxostat, and potential regulatory actions.

6 of 169 This is a brief introduction to the main issues for discussion, which are described in more detail in the attached reviews.

Background Febuxostat is a xanthine oxidase inhibitor that was approved for marketing in the United States in February 2009 following three review cycles. Concerns related to CV safety and death were the basis for FDA not approving febuxostat on the first two review cycles. Prior to approval there was an AAC meeting on November 24, 2008, which focused on the drug’s CV safety. Additional data from a new clinical trial submitted in the third review cycle were determined to be sufficient to support approval. Upon approval, a post-marketing safety trial was required to evaluate the CV safety of febuxostat. This safety trial, the CARES study, has recently been completed and is the reason for this AC meeting.

Gout Gout is a metabolic disorder characterized by reduced clearance or overproduction of uric acid, leading to hyperuricemia, which can result in monosodium urate (MSU) crystal formation and deposition around the joints and soft tissues, urate nephropathy, and nephrolithiasis. The prevalence of gout has been increasing over the past few decades and has been recently estimated to affect approximately 3.9% of adults in the United States (8.3 million)1, surpassing rheumatoid arthritis as the most common form of inflammatory arthritis. The condition affects primarily middle-aged and older men and post-menopausal women. Obesity, hyperlipidemia, diabetes, hypertension, chronic renal kidney disease, metabolic syndrome, and CV disease are frequent comorbidities in patients with gout.

Available Therapies It is important to understand the available therapies for the treatment of gout for context as you consider the benefit-risk assessment of febuxostat. Treatment and prevention of gout involves anti-inflammatory medications for acute attacks and urate lowering therapies for management of hyperuricemia.

Treatment of acute attacks utilizes anti-inflammatory treatment, such as colchicine, nonsteroidal anti-inflammatory drugs (NSAIDS), or corticosteroids. When urate-lowering therapies are initiated, patients are at an increased risk of gout attacks, so it is common practice to use an agent to help reduce the frequency and severity of acute gout attacks. Maintenance doses of either colchicine or an NSAID are continued as prophylaxis against gout flares, typically until the serum uric acid level has been maintained within the target range and there have been no acute attacks for 3 to 6 months.

Control of hyperuricemia is the foundation of management of gout. Urate-lowering therapies include the following:

1 Zhu Y, Pandya BJ, Choi HK, “Prevalence of gout and hyperuricemia in the US general population: the National Health and Nutrition Examination Survey 2007-2008.” Arthritis Rheum 2011; 63:3136-3141.

7 of 169 • Xanthine oxidase inhibitors (XOI) - allopurinol and febuxostat • Uricosuric therapies - probenecid, sulfinpyrazone, and lesinurad • Uricase products – pegloticase

XOI lower uric acid production and are considered first-line therapies.2 Allopurinol has been available since the 1960s and is now a generic product. Febuxostat was approved in the United States in 2009. Based upon the analysis of drug utilization from the FDA Office of Surveillance and Epidemiology (OSE), allopurinol is the most widely used urate-lowering therapy in the United States with 14.9 million prescriptions dispensed, compared to 1.2 million prescriptions for febuxostat in 2017. Refer to the OSE review for additional details on the utilization of febuxostat in the United States.

Allopurinol does have known safety issues, including serious skin reactions and hepatotoxicity.3 Patients positive for the HLA -B*5801 allele (estimated to be up to 7% of all races but most commonly found in patients of Korean, Han Chinese, and Thai descent) are predisposed for developing allopurinol hypersensitivity syndrome (AHS).4 Febuxostat also has safety issues related to hepatotoxicity and serious skin reactions.5 Based upon post-marketing reports, many patients with serious skin reactions to febuxostat have a history of skin reactions to allopurinol.

Uricosuric therapies increase urinary acid excretion and are considered second line therapies. There are known limitations with these products, particularly with respect to use in patients with urolithiasis or renal impairment. Finally, uricases break down uric acid. Because humans do not possess endogenous uricase, uricase products are highly immunogenic resulting in loss of efficacy and safety issues, such as hypersensitivity reactions and anaphylaxis. Because of this, these products can only be used for limited periods of time. These products are reserved for patients with severe tophaceous gout who are refractory to conventional therapy.6

Overall, the number of therapies for gout is limited and there remains an unmet need for treatments for this disease. As you consider the results of CARES and the benefit-risk assessment of febuxostat, it will be important to consider other available therapies for gout. The Clinical Review provides a more detailed discussion of the available therapies for gout to help inform your benefit-risk discussion.

2 Khanna D et al, 2012 American College of Rheumatology Guidelines for Management of Gout. Part 1: Systematic Nonpharmacologic and Pharmacologic Therapeutic Approaches to Hyperuricemia, Arthritis Care & Research Vol. 64, No. 10, October 2012, pp 1431–1446. 3 FDA-approved allopurinol labeling 4 Zineh I, Mummameni P, Lyndly J, La Grenade LA, Chang SH, et al. Allopurinol pharmacogenetics: assessment of potential clinical usefulness. Pharmacogenomics 2011:12:1741-9. 5 FDA-approved febuxostat labeling 6 FDA-approved pegloticase labeling

8 of 169 Regulatory History of Febuxostat To support approval of febuxostat, the applicant conducted a clinical program consisting of several pivotal controlled clinical studies in patients with gout that included an active comparator. The active comparator in both studies was the other marketed XOI, allopurinol. The primary efficacy evaluation was based upon the surrogate endpoint of serum uric acid (sUA) <6 mg/dL, which is considered an acceptable endpoint for gout development programs. Maintenance of sUA less than 6 mg/dL is a target for the treatment of gout. The incidence of gouty attacks and reduction in tophi (both size and number) were also evaluated as secondary endpoints in these studies.

The original NDA for febuxostat was submitted on December 14, 2004, and proposed febuxostat doses of 80 mg and 120 mg once daily. During the first NDA review cycle, the efficacy of febuxostat in lowering serum uric acid was not at question. The original NDA for febuxostat received a Complete Response (CR) action based upon concerns regarding a numerical increase in CV adverse events and all cause death. For the second cycle, the applicant conducted additional analyses to address FDA’s safety concerns; however, FDA again issued a CR action based upon continued concerns regarding cardiovascular safety. FDA recommended that the applicant conduct a new clinical trial to further evaluate the cardiovascular safety of febuxostat. Following the second CR action, the applicant conducted a new clinical trial that was larger and had pre-specified assessment of CV endpoints and adjudication – CONFIRMS (F-GT06-53) – a randomized, double-blind, controlled trial comparing allopurinol, febuxostat 80 mg and a lower dose of febuxostat 40 mg. For the third cycle, the applicant submitted the results of CONFIRMS.

The results of CONFIRMS did not show an increased risk of CV death with febuxostat compared to allopurinol. The febuxostat NDA was discussed at an Arthritis Advisory Committee meeting on November 24, 2008. The AC panel voted 12 to 0 with 1 abstention, to recommend approval of febuxostat with a post-marketing safety study to further assess the CV risk associated with the drug. Febuxostat was approved on February 13, 2009. In the approval letter, FDA issued the following post-marketing requirement (PMR):

A randomized, controlled trial of adequate size and duration to determine whether the use of febuxostat is associated with a moderate increase in the risk of serious adverse cardiovascular outcomes as compared to allopurinol Final Protocol Submission: August 2009 Study Completion January 2014 Final Report Submission January 2015

Due to challenges with enrollment, the milestones were modified with a Final Report Submission date of May 2019.

9 of 169 CARES Study The CARES Study (Study TMX-67-301) was a multicenter, randomized, double-blind, double- dummy, active-controlled, CV outcome safety study conducted in the U.S., Canada, and Mexico. The study was non-inferiority in design. The primary endpoint was a composite of major adverse CV events (MACE): CV death, nonfatal MI, nonfatal stroke and unstable angina with urgent coronary revascularization. Numerous secondary endpoints included the individual components of the MACE composite as well as death from any cause. All suspected cardiovascular events were prospectively adjudicated by an independent Cardiovascular Endpoint Committee (CEC) blinded to treatment assignment.

As per treatment guidelines for chronic gout, doses of both urate lowering therapies were titrated to a maximum dose of 80 mg QD febuxostat and 600 mg QD allopurinol based on a goal of achieving a sUA level <6 mg/dL and on renal function. During the first 6 months of the study, subjects received prophylactic gout therapy with either colchicine, naproxen or another NSAID, or prednisone. Although the protocol permitted the short-term use (<4 weeks) of NSAIDs and COX-2 inhibitors, long-term (>4 weeks) use of these drugs or high dose corticosteroids (>10 mg/day of prednisone or equivalent) other than protocol required prophylactic therapy supplied by the applicant was not allowed.

The study was designed to accrue 624 MACE events and have 90% power to rule out a hazard ratio risk margin for MACE (febuxostat vs. allopurinol) greater than 1.3 at one-sided 2.5% alpha level, assuming a true hazard ratio of 1.0. The duration of this study was dependent on the number of MACE events and was projected to be approximately 9 years with a variable duration of treatment for each patient. It was initiated in April 2010 and was completed in July 2017 after 624 MACE events had occurred in 6,198 patients enrolled in the double-blind treatment period of the study. Patients who discontinued study medication were followed for events of interest (MACE); however, there were still a number of patients who were lost to follow up. The applicant engaged a search company to locate patients lost to follow up to minimize the missing data.

In November 2016, the Applicant informed the FDA that the Data Monitoring Committee (DMC) had noted an increase in all-cause mortality (due to cardiovascular mortality) in the CARES study. The study was fully enrolled, but it hadn’t reached the target number of events yet. The DMC had recommended the study continue without modification. FDA advised the applicant that decisions regarding the study conduct were the responsibility of the DMC and applicant; FDA did not want to interfere with the conduct of the study. In March 2017, the applicant informed the FDA that they were allowing the study to continue to reach the targeted number of MACE.

In October 2017, Takeda informed FDA of the results of the primary analysis from CARES, which showed a significant increase in CV death and all-cause death in febuxostat compared with allopurinol. They noted plans to submit the study report in January 2018. On November 15,

10 of 169 2017, FDA issued a Drug Safety Communication about the risk of CV death and all-cause death based upon preliminary information.7 The final study report and supplement were submitted January 19, 2018. The applicant has proposed labeling changes adding information from CARES to the existing Warning and Precaution regarding cardiovascular events in the febuxostat product label.

On June 21, 2018, Public Citizen submitted a Citizen’s Petition requesting the immediate removal of febuxostat from the US market based on the results from the CARES study.

CARES Results A total of 6,190 subjects with a history of gout, hyperuricemia, and major CV disease were enrolled. Approximately 57% of randomized subjects discontinued study drug and 45% of subjects discontinued the study visits. Subject disposition was comparable between treatment arms: the median on-study follow-up time was 2.6 years and the median on-treatment follow- up was 2.0 years. The maximum subject follow-up was 7 years. Due to the high trial discontinuation rate, at the end of trial, the applicant updated the vital status information based on public databases for subjects who were lost to follow up or who terminated the trial early. Of the 6190 subjects randomized in the trial, the last known vital status was unknown for 821 (13%) subjects.

Results of the pre-specified primary analysis of MACE based on a stratified Cox proportional hazards model are shown in the table below. A total of 656 MACE events were observed during the trial in the primary on study analysis. The estimated hazard ratio (HR) of MACE associated with febuxostat relative to allopurinol was 1.03 with a nominal 95% confidence interval of (0.89, 1.21), excluding the pre-specified risk margin of 1.3. While the results for the MACE composite excluded the pre-specified risk margin, there was an increased risk of cardiovascular death, with a HR of 1.34 (1.03, 1.73).

Table 1 Primary Analysis of MACE – On Study Analysis

Febuxostat Allopurinol N=3098 N=3092 Hazard Ratio* PY=8799.5 PY=8675.7 (95% CI) MACE 335 [3.8] 321 [3.7] 1.03 (0.89, 1.21) Cardiovascular (CV) Death 134 100 1.34 (1.03, 1.73) Non-fatal MI 111 118 0.93 (0.72, 1.21) Non-fatal Stroke 71 70 1.01 (0.73, 1.41) Unstable Angina with Urgent Coronary Revascularization 49 56 0.86 (0.59, 1.26) *HR for febuxostat vs. allopurinol PY=person-year [] indicates incidence rate per 100 person-years

7 https://www.fda.gov/Drugs/DrugSafety/ucm584702.htm

11 of 169 Source: Created by the statistical reviewer

For CVOT and other safety trials, FDA generally considers the on-study analysis as the primary analysis of interest. This analysis includes available data on patients throughout the study even after discontinuation of study medication. For example, if a patient discontinued study medication after 2 years of treatment and experienced an MI at year 3 while still in the CARES study, the MI at year 3 is included in the analysis. The on-study analysis maintains the comparability of the treatment arms obtained through the initial randomization and avoids potential biases due to noncompliance, unequal discontinuation, and potential unmeasured post-treatment confounders.

In order to explore the robustness of the on-study analyses and the impact of treatment discontinuation on the primary analysis, the FDA also evaluated results for CV death using different windows of time after treatment discontinuation. Results are summarized in Table 2. These time to event analyses were based on the same stratified Cox proportional hazards model as the primary analysis. There are a couple points to note about these additional analyses. Although a total of 234 CV deaths were observed during the trial (on-study), only 37 CV deaths (16% of CV deaths) occurred on-treatment. Thus, the majority of CV deaths (84%) occurred off study treatment. While the HR was not statistically significant for the on- treatment study populations, when looking at CV deaths on-treatment + 30 or 60 days, a significant increase in CV deaths in the febuxostat arm was noted.

Table 2 Analysis of CV Death and All Cause Death in CARES

Febuxostat Allopurinol Hazard Ratio N=3098 N=3092 (95% CI) CV Death (on-study) 134 100 1.34 (1.03, 1.73) CV Death (on-treatment) 23 14 1.62 (0.84, 3.16) CV Death (on-treatment + 30 days) 62 41 1.49 (1.01, 2.22) CV Death (on-treatment + 60 days) 76 53 1.42 (1.00, 2.01) All-Cause Death (on-study) 243 199 1.22 (1.01, 1.47) All-Cause Death (on-treatment) 36 27 1.31 (0.80, 2.16) All-Cause Death (on-treatment + 30 days) 92 72 1.26 (0.93, 1.72) All-Cause Death (on-treatment + 60 days) 119 96 1.22 (0.94, 1.60) *HR for Febuxostat vs. Allopurinol Source: Created by the statistical reviewer

Table 2 also includes the results for all-cause death. Results shows a significant increase in all cause death (on study) for febuxostat vs. allopurinol with a HR of 1.22 (1.01., 1.47), but results were not significant for other treatment discontinuation windows assessed. The increase in risk of all-cause death was primarily driven by the imbalance of sudden deaths adjudicated as CV deaths.

The cause of the increased risk of CV is unclear as there was no apparent CV event driving the CV death. Sudden death would often be considered secondary to coronary heart disease or

12 of 169 arrhythmias, but we did not see an increased risk of non-fatal MI or other CV events, such as arrhythmias, with febuxostat. The Division of Cardiorenal Products (DCRP) consult team concluded that the data are not compelling for an increased risk of CV death with febuxostat. Refer to the Division of Cardiorenal Products review for details.

In pre-specified subgroup analyses, NSAID use and aspirin use, had a nominally significant statistical interaction with treatment on the risk of CV death. The estimated relative risk of CV death associated with febuxostat was higher among subgroups of NSAID users and aspirin non- users. The significance of these findings is unclear.

The estimated number needed to harm (NNH) for CV death associated with febuxostat was 278, which means that 278 patient-years of exposure to febuxostat are needed to observe one additional CV death relative to allopurinol. The NNH for all-cause death was 217.

Other Data Sources The FDA’s Office of Surveillance and Epidemiology (OSE) reviewed published literature as part of a comprehensive evaluation of the cardiovascular safety of febuxostat. The review focused on a retrospective cohort study comparing febuxostat and allopurinol in patients with gout in the Medicare population by Zhang et al.8 This study assessed hospitalization for MI or stroke, and all-cause mortality, as well as heart failure and coronary revascularization. The authors concluded there was no difference in the risk of MI, stroke, coronary revascularization, heart failure, or all-cause mortality between febuxostat and allopurinol initiators. However, there was a trend toward an increased risk of all-cause mortality in long-term users of febuxostat (>3 years). Refer to the OSE review for a detailed review of this study as there are design considerations with this study.

OSE also conducted an analysis of the utilization of febuxostat in the Sentinel Distributed Database (SDD) to understand the characteristics of the patient population that is prescribed febuxostat in a real-world setting. Analysis of data from the SDD showed that there are differences between patients who use febuxostat in the real-world compared to the patients in the CARES trial. Patients in the CARES trial had more cardiovascular disease and chronic kidney disease than patients in the SDD. For example, 39% to 40% of patients in CARES had a history of MI compared to 1.5% of allopurinol and febuxostat users in the SDD. This is not surprising given that CARES was enriched for patients with CV disease. However, it is important to keep this in mind when considering the generalizability of the results of CARES to the patient population that uses febuxostat in the real world. In addition, in CARES, patients tended to be younger; 48.9% and 51.3% were 65+ years of age in the febuxostat and allopurinol groups respectively, compared to 66% and 64.2 % in the SDD. In CARES, patients remained on study medication for a median duration of exposure over 700 days; however, in the SDD, median duration of exposure to febuxostat and allopurinol was 210 and 334 days, respectively.

8 Zhang M, Solomon DH, Desai RJ, et al. Assessment of Cardiovascular Risk in Older Patients with Gout Initiating Febuxostat versus Allopurinol: A Population-Based Cohort Study. Circulation 2018 Sep 11; 138(1): 1116-1126.

13 of 169

Of note, the applicant is conducting a prospective, randomized, controlled, open, blinded endpoint, CV outcomes post-marketing study (FAST: Febuxostat versus Allopurinol Streamlined Trial) in the United Kingdom and Denmark at the request of the EMA.9 This study is in patients 60 years of age and older with gout who are prescribed allopurinol. After an allopurinol run-in phase, patients are randomized to allopurinol or febuxostat. The primary endpoint is the first occurrence of the Anti-Platelet Trialists’ Collaboration (APTC) cardiovascular endpoint of non- fatal MI, non-fatal stroke, or CV death. The primary analysis is a non-inferiority analysis with a pre-specified upper limit margin for the HR of 1.3. According to the applicant, this study has completed enrollment with a total of 6,142 subjects randomized and has accrued a total of 456 targeted MACE events as of January 2018. However, key results from the FAST study will not be available until May 2020.

There is no clear mechanism of action for an increased risk of cardiovascular death with febuxostat as compared with allopurinol. Preclinical studies did not suggest an increase in CV risk with febuxostat.

Risk Summary FDA required CARES because of concerns related to CV events and death in the original febuxostat development program. The design and conduct of CARES are considered adequate. Results for the primary MACE composite endpoint excluded the pre-specified NI margin for MACE; however, results showed a significant increased risk of CV death, with a HR of 1.34 (1.03, 1.73). The cause of CV death or mechanism of action for this increased risk are not clear. In addition, there are considerations regarding the generalizability of the results.

We ask for your discussion of the results of CARES and the strength of the findings for CV mortality based upon the available data.

Benefits of Febuxostat In order to discuss the benefit-risk assessment for febuxostat, the benefits of febuxostat need to be considered. In the original clinical development program, febuxostat reduced sUA. In general, a higher proportion of patients on febuxostat achieved the target sUA of 6 mg/dL or below compared to allopurinol at doses up to 300 mg daily. Febuxostat did not show a reduction in gouty flares or tophi. Efficacy was assessed in CARES and the results are largely consistent with the original clinical development program. Results showed a numerically larger proportion of patients achieving target sUA of < 6 mg/dL on febuxostat compared to allopurinol. However, there was no difference in gouty flares or tophi resolution with febuxostat compared to allopurinol.

9 MacDonald TM, Ford I, Nuki G, et al. Protocol of the Febuxostat versus Allopurinol Streamlined Trial (FAST): a large prospective, randomized, open, blinded endpoint study comparing the cardiovascular safety of allopurinol and febuxostat in the management of symptomatic hyperuricaemia. BMJ Open 2014; 4:3005354.

14 of 169 There are a limited number of ULTs available in the face of the growing gout epidemic, and an unmet medical need in the gout population. Febuxostat is the only xanthine oxidase inhibitor alternative to allopurinol. This may be important for patients who cannot tolerate allopurinol.

We ask for you to discuss the benefits of febuxostat to help frame the benefit-risk considerations.

Benefit Risk Considerations The purpose of this Advisory Committee meeting is for you to discuss the results of CARES, discuss the benefit –risk assessment for febuxostat and potential regulatory actions. The first issue we want you to discuss is the strength of the findings for CV mortality based upon the available data. Important considerations for this discussion include the following: paradoxical finding of exclusion of a significant increase in MACE with an increased risk of CV death, questions regarding generalizability, and unclear biologic plausibility. You will also be asked to discuss the benefits of febuxostat. When discussing this issue, it is important to consider other available therapies for gout and the current landscape of gout management.

Following that discussion, we ask for you to discuss whether the benefit–risk profile of febuxostat has changed. If you consider the benefit –risk profile of febuxostat to have changed, we also ask for you to describe patient populations in which the benefits of use of febuxostat outweigh the risks and patient populations in which the benefits of use of febuxostat do not outweigh the risks.

When faced with new safety information, FDA has a number of regulatory options that we ask the panel to consider for their discussion, as there are multiple approaches the Agency may undertake to mitigate risk. Generally, the first consideration for risk mitigation is labeling. Labeling options include a Boxed Warning, Contraindications, Warnings and Precautions, Limitations of Use and Medication Guide (FDA required patient labeling). A Risk Evaluation and Mitigation Strategy (REMS) is also a regulatory consideration. However, the Agency’s approach is generally to consider whether the risks can be adequately communicated with revised labeling including a boxed warning, concise indications, and a Medication Guide for patients. A REMS might be considered, if it can reasonably mitigate the risk and ensure the benefits of the drug outweigh its risk. In the Agency’s experience, REMS programs have minimal impact in mitigating the risk of MACE or CV death unless the REMS is designed specifically to limit who might receive the product. It is important to keep in mind that such a REMS may not have the intended impact and is likely to require elements which could be burdensome to healthcare providers and patients. Therefore, we ask the panel to consider the following regulatory considerations and importantly, their potential clinical and patient care impact: • Add the results of CARES to the febuxostat label o e.g. updating the existing Warning and Precaution regarding cardiovascular events • Add a boxed warning to the febuxostat label o A boxed warning can be used to highlight an adverse reaction that is so serious in proportion to the potential benefit from the drug that it is essential that it be considered in assessing the risks and benefits of using the drug.

15 of 169 • Modify the labeling to limit use of febuxostat to second line therapy o This could be accomplished by a change in the indication statement and/or a limitation of use. • Withdraw febuxostat from the market o On June 21, 2018, Public Citizen submitted a Citizen’s Petition requesting the immediate removal of febuxostat from the US market based on the results from the CARES study

We thank you for your participation in this important Advisory Committee meeting. We look forward to the discussion.

16 of 169 3 Draft Points to Consider

1. Discuss the results of the “Cardiovascular Safety of Febuxostat and Allopurinol in Patients with Gout and Cardiovascular Morbidities (CARES)” study, particularly major adverse cardiovascular events (MACE) and cardiovascular (CV) mortality. Please consider the following in your discussion: a. Biologic plausibility of CV mortality b. Strength of the findings for CV mortality, considering the totality of available data

2. Discuss the benefits of febuxostat for the treatment of hyperuricemia in patients with gout.

3. Given the results of the CARES study, discuss whether the benefit-risk profile of febuxostat for the treatment of hyperuricemia in patients with gout has changed. Address the following in your discussion: a. Discuss any patient populations in which the benefits outweigh the risks of the use of febuxostat. b. Discuss any patient populations in which the benefits do not outweigh the risks of the use of febuxostat.

4. Discuss the following potential regulatory activities in response to the results of the CARES study and the potential clinical impact of these options. a. Update existing warning regarding Cardiovascular Events in the febuxostat product label b. Addition of a boxed warning to the febuxostat product label c. Modify labeling to limit use of febuxostat to second line therapy (e.g. 2nd line therapy in patients who have failed allopurinol) d. Withdrawal febuxostat from the market

5. Based upon the available data, is there a patient population in which the benefit-risk profile for febuxostat is favorable for the treatment of hyperuricemia in patients with gout?

a. If yes, describe the patient population with a favorable benefit-risk profile for use of febuxostat and any other recommendations (e.g. labeling changes). b. If no, discuss your rationale, the impact of this recommendation, and any other recommendations you may have.

17 of 169 4 Clinical and Statistical Review

ARTHRITIS ADVISORY COMMITTEE AND DRUG SAFETY AND RISK MANAGEMENT ADVISORY COMMITTEE MEETING January 11, 2019

Clinical and Statistical Review

NDA 21856 Febuxostat Xanthine oxidase (XO) inhibitor for Gout

Takeda

18 of 169 4.1. Executive Summary

Febuxostat (Uloric) is a xanthine oxidase inhibitor approved by the FDA in February 2009 for the chronic management of hyperuricemia in patients with gout as first line therapy for this disease. There were 3 review cycles before the FDA approval of febuxostat. Concerns related to CV safety and death were the basis for FDA not approving febuxostat on the first two review cycles. Upon approval, a post-marketing safety trial was required to evaluate the CV safety of febuxostat.

The applicant submitted the results from the cardiovascular outcomes study TMX-67-301 (CARES trial) to fulfill a post-marketing required safety study. This study was a multicenter, double-blind, noninferiority trial comparing febuxostat to allopurinol in 6190 patients with gout and significant cardiovascular disease. As per treat-to-target guidelines for the management of chronic gout, doses of both urate lowering drugs were titrated to a maximum dose of 80 mg QD febuxostat and 600 mg QD allopurinol based on a goal of achieving a sUA level <6 mg/dL and on renal function. Study subjects also received prophylactic gout therapy with either colchicine, an NSAID, or prednisone during the first six months of the trial.

The primary endpoint was a composite of the major adverse cardiovascular events (MACE). Numerous secondary endpoints included the individual components of the MACE composite as well as death from any cause. The trial was designed to accrue 624 MACE events and have 90% power to rule out a hazard ratio risk margin for MACE (febuxostat vs. allopurinol) greater than 1.3 at one-sided 2.5% alpha level, assuming a true hazard ratio of 1.0. A total of 656 MACE events (335 with febuxostat and 321 with allopurinol) were observed during the trial. The estimated hazard ratio of MACE associated with febuxostat relative to allopurinol was 1.03 with a nominal 95% confidence interval of (0.89, 1.21), which excluded the pre-specified risk margin of 1.3.

Prespecified secondary analyses showed an increased risk for all cause death (HR 1.22 [95% CI 1.01, 1.47]). This increase in risk for all-cause death was primarily driven by an increased risk of CV death (HR 1.34 [95% CI 1.03, 1.73]) associated with febuxostat. These results were supported by sensitivity analyses. In pre-specified subgroup analyses, NSAID use and aspirin use had a nominally significant statistical interaction with treatment on the risk of CV death. The estimated relative risk of CV death associated with febuxostat was higher among subgroups of NSAID users and aspirin non-users. No subgroup population or baseline characteristic associated with an increased risk for CV death with febuxostat was identified.

Results from the analyses of gout endpoints (proportion of patients achieving serum uric acid [sUA] <6 mg/dL, rate of gout flares and tophi resolution) following the completion of the dose titration process were generally similar between the two treatment groups.

A total of 442 deaths occurred in this study, out of which 234 deaths were adjudicated as CV deaths. The majority of CV deaths (197 of 234; 84%) occurred after discontinuation of study medication while only 37 CV deaths (16%) occurred while subjects were taking their study

19 of 169 medication. Rates for all-cause death and adjudicated CV death were higher with febuxostat (243 of 3,098 [7.8%] and 134 of 3,098 [4.3%], respectively) compared to allopurinol. Sudden cardiac death was the most common cause of adjudicated CV deaths in the febuxostat group (83 of 3,098; 2.8%) as compared to allopurinol (56 of 3,092; 1.8%). Based on the clinical review of the overall safety for this study, no other new safety signals or concerns were identified. A consultative review of the CV safety results by the Division of Cardiorenal Products concluded that the data were not compelling for an increased risk of CV death with febuxostat and the previously reviewed non-clinical data did not identify potential mechanisms/causes for this finding.

The submission of the results from the post-marketing-required cardiovascular outcomes study TMX-67-301 raises questions whether the risk-benefit assessment for febuxostat may have changed as a result of the persistent safety questions regarding the drug’s CV safety profile. Overall, study TMX-67-301 was well-designed to generate the CV safety data to address the outstanding post-marketing requirement to determine whether the use of febuxostat is associated with a moderate increase in the risk of serious adverse cardiovascular outcomes as compared to allopurinol. However, results from this trial also provide evidence for safety concerns that were identified by the original clinical team regarding the increase in all-cause mortality and cardiovascular mortality with febuxostat. These safety concerns were noted during the first review cycle submitted in support of the drug’s overall risk-benefit assessment and marketing approval. Limitations on the results from TMX-67- 301 include the generalizability of these results to the broader gout population since, by design, the subjects evaluated in this study were at higher risk for cardiovascular disease than patients who may use febuxostat in a real-world setting.

Given that febuxostat is the only XOI alternative to allopurinol, the limited number of urate lowering therapies (ULT) that are available (which all have their own inherent safety risks), and the growing gout epidemic, there is an unmet medical need for safe and efficacious urate lowering therapies. The advisory committee panel will be asked to discuss the cardiovascular safety findings from TMX-67-301, and whether the risk-benefit profile of febuxostat has changed, and what risk mitigation actions (if any) they suggest the FDA should put into place to ensure the drug’s safe use.

20 of 169 4.2. Therapeutic Context

4.2.1. Analysis of Condition

Gout is a metabolic disorder characterized by reduced clearance or overproduction of uric acid leading to hyperuricemia, which in turn can result in monosodium urate (MSU) crystal formation around the joints and soft tissues, urate nephropathy, and nephrolithiasis. The prevalence of gout has been increasing over the past few decades and has been recently estimated to affect approximately 3.9% of adults in the United States (8.3 million)10. The condition affects primarily middle-aged and older men and post-menopausal women. Obesity, hyperlipidemia, diabetes, hypertension, chronic renal insufficiency, metabolic syndrome, and cardiovascular disease are frequent comorbidities in patients with gout.

The course of gout is characterized by acute attacks of gouty arthritis alternating with attack- free periods of intercritical gout. A typical course of gouty arthritis attack (or gout flare) is characterized by acute inflammation of the affected joint and surrounding tissues associated with often excruciating pain, tenderness, erythema, and swelling. If left untreated, the acute inflammatory episode is self-limited, typically peaking within 24-48 hours and eventually subsiding within 7-10 days. Treatment of acute attacks utilizes anti-inflammatory treatment of various mechanisms, such as colchicine, nonsteroidal anti-inflammatory drugs (NSAIDS), or corticosteroids. It is common practice to use an agent to help reduce the frequency and severity of acute gout attacks, for which a patient is at increased risk during initiation of uric- acid lowering therapies. To this end, maintenance doses of either colchicine or an NSAID are continued as prophylaxis against gout flares, typically until the serum uric acid level has been maintained within the target range and there have been no acute attacks for 3 to 6 months.

The chronic management of gout is founded upon control of hyperuricemia, as only this approach treats the underlying pathology of the disorder. The mechanistic approaches to lowering serum uric acid (sUA) include: • Lowering uric acid production. This is currently the most common approach to treatment, via xanthine oxidase inhibitors, i.e., allopurinol and febuxostat. • Increasing urinary uric acid excretion (uricosurics). Uricosurics such as probenecid and lesinurad inhibit active renal reabsorption of uric acid through urate transporters in proximal tubule epithelial cells (predominantly URAT1), resulting in increased urinary uric acid excretion. • Direct enzymatic breakdown of uric acid. Because humans do not possess an endogenous uricase, drugs such as pegloticase and rasburicase are derived from foreign

10 Zhu Y, Pandya BJ, Choi HK, “Prevalence of gout and hyperuricemia in the US general population: the National Health and Nutrition Examination Survey 2007-2008.” Arthritis Rheum 2011; 63:3136-3141.

21 of 169 proteins, and their use is limited by immunogenicity. Uricase breaks down uric acid into the much more soluble allantoin, which can then be excreted in the urine.

4.2.2. Analysis of Current Treatment Options

Table 3 lists the currently approved small molecule products as well as the therapeutic biologic treatments for the management of hyperuricemia.

Table 3: Treatments for the Management of Hyperuricemia

Product Year of Approval Indication

Xanthine Oxidase Inhibitors (XOIs) Management of patients with signs and symptoms of Allopurinol 1966 primary or secondary gout (i.e., acute attacks, tophi, joint destruction, uric acid lithiasis, and/or nephropathy) Chronic management of hyperuricemia in patients with Febuxostat 2009 gout Uricosuric Agents1 Treatment of the hyperuricemia associated with gout Probenecid 1951 and gouty arthritis 1959 Treatment of chronic gouty arthritis and intermittent Sulfinpyrazone (Removed from gouty arthritis market 2002) Treatment of hyperuricemia associated with gout in Lesinurad 2015 patients who have not achieved target serum uric acid (sUA) levels with a xanthine oxidase inhibitor alone Combined Uricosuric Agent and XOI Lesinurad/Allopurinol Treatment of hyperuricemia associated with gout in FDC patients who have not achieved target serum uric acid 2017 (sUA) levels with a medically appropriate daily dose of allopurinol alone Uricase Initial management of plasma uric acid levels in pediatric patients with leukemia, lymphoma, and sold tumor Rasburicase 2002 malignancies who are receiving anti-cancer therapy expected to result in tumor lysis and subsequent elevation of plasma uric acid Treatment of chronic gout in adult patients refractory to Pegloticase 2010 conventional therapy 1Benzbromarone is a uricosuric agent that was never marketed in the U.S. but is available in other countries

Benzbromarone, sulfinpyrazone, probenecid and lesinurad comprise the uricosuric class of drugs which can be used in patients who are underexcretors of urate. Probenecid and lesinurad are the only uricosuric currently available in the U.S. Like lesinurad, probenecid also interferes

22 of 169 with renal absorption of uric acid by inhibiting URAT1. Probenecid also inhibits OAT1 and OAT3, as well as GLUT9. Based in part on concerns regarding urolithiasis and decreased efficacy in patients with creatinine clearance below 50 ml/min, American College of Rheumatology treatment guidelines include caveats, such as not using uricosurics in patients with urolithiasis or in patients with a creatinine clearance below 50 ml/min, or in patients with elevated urine uric acid. The guidelines also recommend monitoring of urinary uric acid during therapy and considering urine alkalinization and increasing fluid intake. In order to minimize the risk of nephrotoxicity associated with lesinurad, patients prescribed this drug also need to take it with a concomitant xanthine oxidase inhibitor.

The xanthine oxidase inhibitors (XOI), allopurinol and febuxostat, are the agents most commonly used as first-line urate lowering therapy in patients with gout and in those with a history of nephrolithiasis (renal stones). The effectiveness of allopurinol is limited by a number of issues including the need to use lower doses in patients with renal insufficiency, and an adverse event profile that includes gastrointestinal, hepatic, renal, hematological and skin toxicities that occur in approximately 20% of patients who take this drug. In addition, hypersensitivity reactions occur in 2-4% of patients that in some instances have been fatal. Patients who are at increased risk for developing severe allopurinol hypersensitivity syndrome (AHS) include a subgroup of the population positive for the HLA -B*5801 allele (estimated to be up to 7% of all races but most commonly found in patients of Korean, Han Chinese and Thai descent).

Febuxostat’s safety profile is generally consistent with that of allopurinol but it does not require renal adjustment in dosing in patients with a creatinine clearance > 30 ml/minute. However, its current label carries warnings for both cardiovascular events and hepatotoxicity some of which have resulted in fatalities. Benzbromarone, sulfinpyrazone and probenecid comprise the uricosuric class of drugs which can be used in patients who are underexcretors of urate. Uricosuric agents are used as second-line therapy since their utility is limited by the magnitude of the treatment effect and the risk for developing urate renal stones and crystalluria in patients who are overexcretors of urate, have decreased renal function (creatinine clearance of <50 mg/minute), and/or are not well hydrated to support good urine flow. Pegloticase is a pegylated formulation of recombinant porcine urate oxidase that is administered intravenously. It is reserved as tertiary therapy as a treatment for patients with severe tophaceous gout who are refractory to conventional therapy. The effectiveness of this therapeutic biologic is limited by the development of neutralizing antibodies and the occurrence of infusion reactions and anaphylaxis which requires patients to be premedicated prior to its administration. Additionally, patients with underlying congestive heart failure have to be monitored for exacerbations post-administration of pegloticase.

23 of 169 4.3. Regulatory Background

4.3.1. U.S. Regulatory Actions and Marketing History

Febuxostat is a xanthine oxidase inhibitor that lowers serum uric acid levels by inhibiting the conversion of xanthine to uric acid. In 2009, febuxostat was approved for marketing in the United States (U.S.) under the brand name Uloric® by Takeda Pharmaceuticals for the chronic management of hyperuricemic patients with gout when administered at dose of 40 mg and 80 mg once a day. It is currently approved in 83 countries worldwide to treat gout and/or hyperuricemia in patients with gout as well as to prevent and treat hyperuricemia in adult patients undergoing chemotherapy for hematologic malignancies at intermediate to high risk of tumor lysis syndrome (TLS) at doses ranging from 60 mg to 120 mg once daily. Outside the U.S. and Canada, febuxostat is marketed by Takeda Pharmaceuticals as Turazive in Mexico, as Adenuric by the Menarini Group in Europe and non-European countries, and as Feburic by Teijin Pharma in Japan.

Febuxostat’s initial label contained Warning and Precaution statements regarding gout flares (mechanistic), an increased risk for cardiovascular events and the occurrence of elevated liver function tests in patients receiving this drug. The drug’s USPI has subsequently undergone four labeling revisions as follows: • 2011: Addition of a contraindication for patients taking xanthine oxidase substrate drugs such as azathioprine or mercaptopurine which could result in elevated drug concentrations and severe toxicity; a corresponding update of Section 7.1 Xanthine Oxidase Substrate Drugs; and updating Section 6.2 Postmarketing Experience to include rhabdomyolysis, psychotic behavior including aggressive thoughts and hypersensitivity reactions including anaphylaxis, Stevens Johnson Syndrome and hypersensitivity skin reactions based on a review of post-marketing case reports • 2012: The Warnings and Precaution statement for elevated liver function tests was updated to hepatic effects and information regarding the occurrence of sometimes fatal hepatic failure was added. A recommendation to do baseline testing and repeat testing of liver function in patients with hepatic symptoms following treatment with febuxostat based on a review of spontaneous post-marketing case reports of hepatic adverse events by the Office of Surveillance and Epidemiology (OSE) was also added. • 2017: Addition of specific dosing recommendations for patients with severe renal impairment (Clcr 15 to 29 ml/min) limiting the dose of febuxostat to 40 mg once day in this population along with revisions to Sections 8.1 Pregnancy and 8.2 Lactation of the label to comply with the Pregnancy and Lactation Labeling Rule (PLLR). • 2018: A Warnings and Precautions statement for the occurrence of serious skin reactions and hypersensitivity reactions including Stevens Johnson Syndrome, drug reaction with eosinophilia and systemic symptoms (DRESS) and toxic epidermal necrolysis (TEN) particularly in patients who had experienced previous similar skin reactions to allopurinol along with updating Section 6.2 Postmarketing Experience to

24 of 169 include these adverse reactions as well as eosinophilia and agranulocytosis based on pharmacovigilance review by OSE.

Review of the European, Canadian, and Mexican labeling for febuxostat reveals that each of these labels contain a warning and precaution for cardiovascular events associated with the drug similar to the current U.S. warning. These countries have also required the marketing authorized holder (MAH) to include additional information such as a recommendation not to use febuxostat in patients with ischemic heart disease or congestive heart failure based on the drug’s phase 3 program (EMA), a potential increase in risk for cardiac failure in patients with pre-existing cardiovascular and/or risk factors for cardiovascular with a recommendation not to administer the drug to patients with ischemic heart disease or congestive heart failure (based on a post-marketing signal in Canada), and a contraindication for use in patients with ischemic cardiomyopathy and congestive heart failure (Mexico). Both the U.S. and the EMA have required the MAH to conduct post-marketing cardiovascular outcome studies to further evaluate this risk.

4.3.2. Summary of Presubmission/Submission Regulatory Activity

Febuxostat was initially approved for marketing in this country in February 2009 following three review cycles that included an Arthritis Advisory Committee (ACC) meeting which focused on the drug’s CV safety.

In support of febuxostat’s safety and efficacy, the applicant initially submitted the results from 2 pivotal, phase 3 studies, C02-009 and C02-010, which were a 6-month active- and placebo- controlled trial and a 12-month, active-controlled trial, respectively. The active comparator in both studies was the other marketed xanthine oxidase inhibitor (XOI), allopurinol, which was administered at doses of 200-300 mg QD pending renal function. Subjects in both trials received colchicine or naproxen as prophylactic therapy for gout flares for the duration of both studies. The primary endpoint evaluated in both trials was the number of subjects who achieved the surrogate endpoint of Serum uric acid (sUA) <6 mg/dL for the last 3 visits. The sUA <6 mg/dL was chosen based on published data which supported this level as the limit of solubility of UA, and the correlation of reduction of the number of gouty attacks and tophi size with reduction of sUA below this level. Theoretically, reduction below this level would allow for UA to be released from body stores, resulting in a net reduction of excess UA. This endpoint had been discussed at the June 2004 Arthritis Advisory Meeting (AAM) who recommended that it be accepted. The incidence of gouty attacks and reduction in tophi (both size and number) were also evaluated as secondary endpoints in these studies.

The original NDA for febuxostat was submitted on December 14, 2004, and proposed febuxostat doses of 80mg and 120mg once daily. At the end of the first review cycle, the clinical review team determined that although febuxostat demonstrated efficacy in lowering sUA levels in patients with gout in a dose dependent manner, it did not show a reduction in gouty flares or tophi, and subjects who had received febuxostat had a higher rate of overall

25 of 169 mortality, mortality from CV causes, and a higher rate of cardiovascular thromboembolic events (Table 4).

Table 4 Summary of Major Safety Results from First Cycle Review of Febuxostat

Placebo Febuxostat Allopurinol (N=172) 40-240 mg QD 100-300 mg QD (N=1707) (N=692) All Deaths1: 0 8 0 CV Deaths2 0 2 0 Ischemic Coronary Artery Disease AEs3 0 17 2 Cardiovascular Thromboembolic AEs4 0 9 0 Phase 2 and 3 studies: TMX—00-004, TMX-01, 005, and C02-009, C02-010 and C02-021 1All Deaths: 3 deaths in subjects who received febuxostat 80 mg QD (retroperitoneal hemorrhage, respiratory failure, and sepsis); 5 deaths in subjects who received febuxostat 120 mg QD (respiratory failure with anoxic encephalopathy, metastatic colon cancer, acute MI, MI, and retroperitoneal hemorrhage) 2CV deaths: one case of acute MI and one case of MI in subjects who received febuxostat 120 mg QD 3Ischemic Coronary Artery Disease AEs: acute coronary syndrome, acute MI, angina, and MI 4Cardiovascular Thromboembolic AEs: strokes and transient ischemic attacks Source: Clinical Safety Review dated 2004

The first approvable letter was subsequently issued on Oct. 14, 2005 citing “concerns regarding the potential for (febuxostat) to cause clinically significant cardiovascular/thrombotic adverse events in excess to that seen with allopurinol or placebo, even when exposure -over-time is factored into the analysis” along with a request for additional information regarding the cardiovascular safety of febuxostat derived from “comparative controlled safety data, or possibly through re-analyses of the current database.”

A complete response submitted by the applicant on February 22, 2006 triggered the second review cycle for febuxostat. To address the agency’s cardiovascular safety concerns, the a applicant submitted a re-analysis and a post hoc adjudication conducted by a single outside cardiologist using the Antiplatelet Trialists’ Collaboration (APTC) criteria of both prior clinical trial safety data as well as new safety data from two ongoing long-term clinical studies C02-021 and TMX-01-005.

As shown in Table 5 below, there were a total of 12 deaths in febuxostat’s NDA safety database, all of which occurred in febuxostat-treated patients. Four out of these 12 febuxostat deaths occurred during the randomized, controlled portion of the phase 3 studies resulting in an all- cause mortality rate of 0.60 deaths/100 patient-years for febuxostat treated subjects versus 0/100 patient-years for allopurinol subjects. The remaining 8 febuxostat deaths occurred during the long-term extension (LTE) studies (all-cause mortality rate of 0.38 deaths/100 patient- years). Overall, the combined all-cause mortality rate for febuxostat was numerically higher (0.43 deaths/100 patient years) compared to the allopurinol treatment group (0 deaths/100 patient years). These imbalances could not be explained by differences in duration of exposure between the febuxostat and allopurinol treatment groups.

26 of 169 Table 5: All-Cause Mortality in Febuxostat Safety Database by Patient-Years of Exposure (Data Cut-off 08 February 2006)

Source: FDA Clinical Safety Review, July 2006

Analyses of investigator-reported primary APTC events also supported a signal. The overall rates per 100 patient-years of exposure for primary APTC events were 1.34 for febuxostat versus 0.3 for allopurinol and 0 for placebo which is consistent with a greater than a 4-fold difference between the febuxostat and allopurinol treatment groups.

Since the blinded adjudication conducted by the applicant’s outside cardiologist of all serious CV AEs observed in the RCTs resulted in a discrepancy in the number of APTC events attributed to febuxostat (7 adjudicated events versus 10 investigator-reported events), the then review division consulted the Division of Cardiovascular and Renal Products (DCRP) for additional input. Due to the lack of sufficient information to apply criteria or otherwise confirm an APTC event, the cardio-renal consultant concluded that adjudication was largely an exercise in futility resulting in acceptance of the investigators’ diagnosis of these serious CV events.

Although the higher rate of death and cardiovascular events in the febuxostat group suggested that the drug may be associated with an increased CV risk, there were multiple limitations in interpreting these data. These limitations included the small number of CV events, a lack of dose response, the limited exposure to allopurinol as a result of unequal randomization in the randomized controlled portions of the studies followed by most patients receiving open-label febuxostat in the LTEs, and the application of APTC criteria and adjudication introduced in a post-hoc fashion, which raised uncertainty about drawing definitive conclusions. In view of these unresolved safety concerns and a benefit assessment based on a surrogate endpoint (lowering serum uric acid <6 mg/dL) that was unsubstantiated by clinically meaningful improvements (no reduction in gout flares or decrease in tophi observed in the RCTs), a second approvable letter was issued by the agency on August 2, 2006. This letter stated that the applicant had to “provide further data to clarify the cardiovascular risks of the proposed doses and/or provide data on the safety and efficacy of lower doses of febuxostat in order to assure us that a dose level(s) with favorable risk-benefit characteristics has been defined.” It was also suggested that the applicant conduct a new study designed to determine whether the cardiovascular safety signal would be seen again and that it would be important that the study

27 of 169 be designed in such a manner as to collect an adequate number of CV AEs to reach conclusions about the CV safety of febuxostat.

To address these persistent CV safety concerns, the applicant submitted in June 2008 efficacy and safety data from a new phase 3 RCT, the CONFIRMS trial (F-GT06-53), as well as an additional 12 months of exposure data from the two LTE studies, TMX-01-005 and C02-021, initiating the third and final review cycle. CONFIRMS was a non-inferiority study specifically designed to address Agency safety concerns and feedback comments as follows: it was larger (2269 subjects) than the previous two phase 3 studies combined (studies C0-009 and C0-010: 1832 subjects); it had three times the number of subjects randomized to the active control arm (allopurinol =765 subjects) than the combined phase 3 studies (combined allopurinol subjects in studies C0-009 and C0-010 = 521); it included approximately 1300 subjects at risk for CV disease, and had pre-specified CV endpoints and a CV adjudication committee.

Unlike the prior phase 3 studies (C0-009 and C0-010), there were numerically more deaths in the CONFIRMS study in subjects exposed to allopurinol (3 deaths: 2 adjudicated CV deaths and 1 death due to necrotizing pneumonia with sepsis S/P lung carcinoma resection) versus febuxostat (2 deaths: 1 adjudicated CV death and 1 death due to brain edema and COPD). Based on pooled, phase 3, RCT data, the exposure-adjusted rate of death for all febuxostat treated subjects was 0.4 deaths/100 patient-years versus 0.5 deaths/100 patient-years in allopurinol treated subjects. Additionally, the exposure-adjusted rate of CV deaths based on the pooled, phase 3 studies was similar between the two treatment groups [all febuxostat treated subjects: 0.1% (3 CV deaths in 2,690 subjects) versus allopurinol treated subjects: 0.2% (2 CV deaths in 1,277 subjects)]. The rate of adjudicated CV APTC events was similar for the febuxostat-treated patients (0.3% in the combined febuxostat groups) versus the allopurinol group (0.4%).

Although no increase in the exposure- adjusted mortality rate was seen based on the pooled data including the LTE studies as compared to the second review cycle, the updated rate of investigator-reported primary APTC events based on pooled data including the LTE studies was 1.2 events/100 patient-years for febuxostat treated subjects versus 0.6 events/100 patient years for the allopurinol group.

Due to persistent safety concerns raised by Agency reviewers, the safety and efficacy of febuxostat was discussed at an Arthritis Advisory Committee meeting held on November 24, 2008. The AC panel voted 12 to 0 with 1 abstention, to recommend approval of febuxostat with a post-marketing safety study to further assess the CV risk associated with the drug. Accordingly, the approval letter issued on February 13, 2009 required the applicant to conduct a “randomized, controlled trial of adequate size and duration to determine whether the use of [febuxostat] is associated with a moderate increase in the risk of serious adverse cardiovascular outcomes as compared to allopurinol” as well as the following warning in the drug’s label:

“Cardiovascular Events: A higher rate of cardiovascular thromboembolic events was observed in patients treated with ULORIC (0.74 per 100 P-Y [95% Confidence Interval (CI) 0.36-1.37]) than

28 of 169 allopurinol (0.60 per 100 P-Y [95% CI 0.16-1.53]). A causal relationship with ULORIC has not been established. Monitor for signs and symptoms of MI and stroke.”

The drug’s approval letter also contained the following timeline milestones for the post- marketing required cardiovascular outcome study: Final Protocol Submission: August 2009 Study Completion January 2014 Final Report Submission January 2015

In July 2009, the FDA initiated discussions with the applicant regarding the overall design of the post-marketing required cardiovascular outcome study for febuxostat. In January 2010, an agreement was reached regarding a final study protocol for study TMX-67-301 (CARES study) that satisfied the requirements outlined in the drug’s approval letter. The applicant submitted semiannual reports containing updates on study enrollment and MACE status to the FDA starting in December 2010. As a result of slow study recruitment and a higher rate of subject discontinuation than had been anticipated during the design of study TMX-67-301, the applicant requested revised milestone dates for this post-marketing required study which were agreed to by the FDA in May 2013 as follows: Study Completion December 2019 Final Report Submission May 2019

In November 2016, the applicant informed the FDA that the Data Monitoring Committee (DMC) had noted an increase in all-cause mortality (due to cardiovascular mortality) in study TMX-67- 301. The study was fully enrolled, but it hadn’t reached the target number of events yet. The DMC had recommended the study continue without modification. Following internal discussions within the agency, the FDA sent the applicant correspondence in December 2016 noting that the FDA will not review the interim data or interfere with the conduct of the study and deferred these decisions to the DMC and the applicant. In March 2017, the applicant informed the FDA that they were allowing the study to continue to reach the targeted number of MACE. In October 2017, the applicant informed the FDA of the results of the primary analysis from study TMX-67-301 which showed a significant increase in cardiovascular death and all-cause death and their plans to submit the study report in January 2018.

On November 15, 2017, FDA issued a Drug Safety Communication (DSC) alerting the public to the increased risk for cardiovascular death and all-cause death associated with febuxostat versus allopurinol based upon preliminary information and noted that the DSC would be updated once the study results had been submitted and a review had been completed. https://www.fda.gov/Drugs/DrugSafety/ucm584702.htm The applicant submitted the final study report for study TMX-67-301 (supplemental NDA 21856/S-013) on January 19, 2018. The results from this trial were subsequently published in NEJM in March 2018.11 On June 21, 2018,

11 White WB, Saag KG, Becker MA, Cardiovascular Safety of Febuxostat or Allopurinol in Patients with Gout, N Engl J Med. 2018 Mar 29;378(13):1200-1210.

29 of 169 Public Citizen submitted a Citizen’s Petition requesting the immediate removal of febuxostat from the US market based on the results from the CARES study. Post hoc analyses of the study were also presented in an abstract at the 2019 American College of Rheumatology Meeting in October 2019.

4.4. Statistical and Clinical Evaluation

4.4.1. Sources of Clinical Data and Review Strategy

Due to imbalances in cardiovascular events not in favor of febuxostat (Uloric) that were observed in early phase 3 studies but were not present in a subsequent confirmatory phase 3 study, the applicant agreed to conduct a post-marketing required (PMR 811-1) cardiovascular safety outcome study as part of the initial approval agreement reached with the Agency for NDA 21856 on February 3, 2009 described as follows:

“A randomized controlled trial of adequate size and duration to determine whether the use of Uloric is associated with a moderate increase in the risk of serious adverse cardiovascular outcomes as compared to allopurinol.”

In fulfillment of this PMR and in support of proposed labeling changes regarding the cardiovascular safety of febuxostat in hyperuricemic patients with gout, the applicant submitted the results from a single, phase 3b study, TMX-67-301, summarized in Table 6. For purposes of the following statistical and clinical review, this study will be hereafter referred to as Study 301.

30 of 169 Table 6: Clinical Trials Relevant to this NDA

Trial Trial Design Regimen/ Study Endpoints Treatment No. of Study Population No. of Identity schedule/ route Duration/ patients Centers and Follow Up enrolled Countries Controlled Studies to Support Cardiovascular Safety TMX-67- Phase 3b Febuxostat: Primary: Screening visit, N= 6190 Males > 50 yo and Total sites = 301 multicenter, 40 mg or 80 mg Time from randomization postmenopausal 320 randomized, orally once daily randomization to the visit and Males= 5196 females > 55 yo with Countries: double-blind, titrated to target first occurrence of monthly visits Females= 994 gout and US = 275, double-dummy, based on sUA <6 any event in the through Month hyperuricemia with Canada = 12, active mg/dL predefined MACE 6, followed by significant Mexico=33 controlled, composite1 visits every 6 cardiovascular co- cardiovascular Allopurinol: months for the morbidities3 outcome study 200 mg to 600 mg Secondary: remainder of orally once daily Time from the study until titrated to target randomization to: first MACE based on renal the first occurrence event function and sUA of any APTC2 event <6mg/dL and the occurrence of each individual event in the pre- defined MACE composite 1Predefined MACE composite endpoint: cardiovascular death, nonfatal myocardial infarction (MI), nonfatal stroke and unstable angina with urgent coronary revascularization. 2Predefined APTC composite endpoint: cardiovascular death, nonfatal myocardial infarction (MI), and nonfatal stroke. 3Cardiovascular comorbidities: myocardial infarction (MI), hospitalized unstable angina, cardiac or cerebrovascular revascularization procedure, stroke, hospitalized transient ischemic attack, peripheral vascular disease or a history of diabetes mellitus with evidence or micro-or macro-vascular disease

31 of 169 Review Strategy This NDA submission contained one randomized trial, study 301, entitled “A Multicenter, Randomized Active-Control, Phase 3B Study to Evaluate the Cardiovascular Safety of Febuxostat and Allopurinol in Subjects With Gout and Cardiovascular Comorbidities” which is also known publicly as the “Cardiovascular Safety of Febuxostat or Allopurinol in Patients with Gout - CARES” study. The statistical and clinical review of cardiovascular safety for study 301 focuses primarily on the time to event analysis of major cardiovascular adverse events (MACE) and includes a detailed review and analysis of data contained in the clinical study report (CSR), case report forms (CRFS), statistical analysis plan (SAP), datasets and SAS program. The clinical review of overall safety is primarily based on the safety population from study 301, which is defined as patients who received at least one dose of study drug and includes a review of safety data contained in the submitted clinical study report, line-listings, CRFs and case narratives.

The statistical and clinical review of efficacy and safety included the following: • Review of the current literature on gout epidemiology and treatment • Review of Study TMX-301, including CSR, protocol, protocol amendments, SAP, and SAP amendments • Review and assessment of applicant analyses of time to CV events and other safety evaluations in the CSR • Review of datasets submitted as SAS transport files • Review of patient narratives for deaths and serious adverse events • Review of minutes and key meeting discussions and written responses regarding the design of the study protocol • Review and assessment of the Module 2 summaries including the Summary of Clinical Efficacy, and Summary of Clinical Safety Integrated Summary of Efficacy, Integrated Summary of Safety, and proposed labeling modifications for febuxostat • Review of consultation report of Office of Scientific Investigations • Request for additional information from the applicant and review of their response • Formulation of the benefit-risk analysis and recommendations • Review and evaluation of proposed labeling

Data Sources

The electronic submission includes protocols, SAP, CSRs, SAS transport datasets in legacy, SDTM, and ADAM format, and SAS codes for the NDA submission. The format, content and documentation of the data submitted in support of this application were adequate to conduct a statistical review of the cardiovascular risk associated with febuxostat based on the Study TMX- 301.

Data and Analysis Quality

Upon further clarifications from the applicant per information request by the FDA, the statistical reviewer was able to:

32 of 169 • Reproduce the applicant’s analysis dataset and analysis results from legacy dataset • Evaluate documentation of data quality control/assurance procedures • Find the blinding and un-blinding procedures • Verify the randomization treatment assignments • Conduct FDA analyses

4.4.2. Review of Relevant Individual Trials Used to Support Efficacy

TMX-67-301: Cardiovascular Safety of Febuxostat and Allopurinol in Patients with Gout and Cardiovascular Morbidities (CARES)

Trial Design and Endpoints

Study 301 was a multicenter, randomized, double-blind, double-dummy, active-controlled, parallel group, non-inferiority trial to evaluate the cardiovascular safety of febuxostat versus allopurinol in gout patients with significant cardiovascular disease. Study candidates taking urate lowering therapy (ULT) were required to discontinue these medications and undergo a 14-day washout before being screened for study eligibility. Additionally, prophylactic gout flare therapy (colchicine 0.6 mg QD, an NSAID or prednisone) was to have been initiated by all subjects during the screening period and continued for the first 6 months of the study. Eligible patients were randomized via a 1:1 ratio stratified by baseline renal function (normal or mild renal impairment [estimated creatinine clearance > 60 mL/min] versus moderate renal impairment [estimated creatinine clearance > 30 but < 60 mL/min) to either treatment with febuxostat 40 mg QD or allopurinol 200 or 300 mg QD. (Determination of the initial daily dose of allopurinol was also dependent on baseline renal function.) Subjects who achieved a target serum uric acid (sUA) level of <6mg/dL at the Week 2 visit continued these initial doses of study medication for the duration of the study. Patients with persistently elevated sUA levels >6.0 mg/dL at the Week 2 visit underwent dose titration of study medications as follows:

• Febuxostat randomized subjects were increased to a dose of 80 mg QD of the drug for the remainder of the trial

• Allopurinol randomized subjects were increased by 100 mg increments monthly of the drug until they achieved either a sUA <6 mg/dL or the maximum daily dose of allopurinol permitted by the protocol which was 400 mg QD for subjects with moderate renal impairment and 600 mg QD for subjects with normal renal function.

To facilitate dose titration of study medications based on sUA response, all subjects’ sUA levels were unblinded to study investigators and the applicant through the completion of the dose titration period (Week 10 visit). All subsequent sUA assessments were blinded to both the investigators and the applicant.

Following randomization, study visits were scheduled at 2, 4, 6, 8, 10, 12, and 24 weeks, after which patients were seen every 6 months for the duration of the trial. Elderly subjects (>65 years old) and patients with moderate renal impairment (estimated creatinine clearance >30

33 of 169 but <60 mL/min) were monitored more frequently due to the risk for xanthine oxidase drug class associated liver function abnormalities with additional study visits scheduled at Months 9 and 15. Since this was an event-driven trial, subjects who had discontinued study treatment but had not withdrawn consent were contacted every 2 months for the duration of the study or until the patient had experienced a cardiovascular adverse event that was positively adjudicated as a major adverse cardiac event (MACE).

All investigators were required to complete a cardiovascular worksheet with relevant clinical information for adjudication of all deaths and potential cardiovascular serious adverse events experienced by subjects over the course of the study. All reported suspected cardiovascular events were prospectively adjudicated by an independent Cardiovascular Endpoint Committee (CEC) blinded to treatment assignment. Interim analyses were performed by an independent statistician following the accrual of approximately 25%, 50% and 75% MACE events. Additionally, a Data Monitoring Committee (DMC) also reviewed unblinded to treatment assignment safety data from the study every 6 months.

The duration of this event-driven study, which was dependent on the occurrence of 624 MACE events in approximately 7500 subjects, was projected to be approximately 9 years with a variable duration of treatment for each subject. It was initiated on 23 April 2010 and was completed on 15 May 2017 after 624 MACE events had occurred in 6198 subjects enrolled in the double-blind treatment period of the study.

34 of 169 Figure 1 Schema of Study 301

Source: Applicant’s Fig. 9.a; p. 27 Clinical Study Report

Eligibility

Table 7 summarizes the major inclusion and exclusion criteria for Study TMX-301. Table 7 – Tabular Summary of Major Inclusion and Exclusion Criteria for Study 301

Major Inclusion Criteria: 1. Males > 50 years old or females > 55 years old and at least 2 years postmenopausal 2. History of (H/O) major cardiovascular or cerebrovascular disease including at least 1 of the following: • Myocardial infarction • Hospitalized unstable angina • Cardiac or cerebrovascular revascularization procedure • Stroke • Hospitalized transient ischemic attach (TIA) • Peripheral vascular disease (ankle brachial index <0.6, revascularization, and/or well- documented H/O claudication) • H/O diabetes mellitus with evidence of microvascular or macrovascular disease (retinopathy, neuropathy, nephropathy, small vessel vascular diseases) 3. Diagnosis of gout as per the American Rheumatism Association Criteria for the Classification of Acute Arthritis of Primary Gout 4. Have either • sUA level > 7.0 mg/dL (> 416 μmol/L) at the Screening Visit OR

35 of 169 • sUA level > 6.0 mg/dL (> 354 μmol/L) at the Screening Visit AND inadequately controlled gout (> 1 flare in the 12 months prior to screening a/or the presence of tophi) Major Exclusion Criteria: 1. Secondary hyperuricemia (e.g., due to myeloproliferative disorder, or organ transplant) 2. H/O xanthinuria 3. Has received urate-lowering therapy (i.e., febuxostat, allopurinol, probenecid, etc.) or excluded medication during the screening period (beginning with Day -14 for subjects on ULT and Day -4) for subjects not taking ULT) 4. H/O hypersensitivity to febuxostat or allopurinol or any components of their formulation; or the subject has a known hypersensitivity to naproxen, any other NSAID, aspirin, colchicine, lansoprazole, or any components in their formulation 5. Has active peptic ulcer disease 6. H/O cancer (other than basal cell carcinoma of the skin) within 5 years prior to the first dose of study medication 7. Had a myocardial infarction or stroke within 60 days prior to the Screening Visit 8. Has an ALT and/or AST values greater than 2 times the upper limit of normal (xULN) during the Screening period 9. Has a significant medical condition and/or conditions that would interfere with the treatment, safety, or compliance with the protocol 10. H/O drug abuse (defined as any illicit drug use) or a H/O alcohol abuse within 5 years prior to the Screening Visit or the subject consumes > 14 alcoholic beverages per week 11. Has received any investigational medicinal product within the 30 days prior to the Screening Visit and throughout the study. In addition, the subject had been previously randomized in this study and received at least one dose of double blind study drug treatment. 12. Subject’s estimated CrCl is <30mL/min, where CrCl is calculated using the Cockcroft and Gault formula based on ideal body weight (IBW) as provided below: Estimated CrCl = (140-age [yr]) x (IBW [kg]) (women multiple by 0.85) 72 x (serum creatinine [mg/dL]) Where IBW was 50 kg for men and 45.5 kg for females, plus 2.3 kg for each inch in height greater than 5 feet (60 inches). 13. H/O infection with hepatitis B or C, or human immunodeficiency virus 14. Subject was required to take a protocol excluded medication

Treatment To ensure maintenance of blind, study medication was supplied as over-encapsulated febuxostat 40 mg, febuxostat 80 mg, allopurinol 200 mg, allopurinol 300 mg and matching placebo capsules that were identical in appearance. The study capsules were packaged in 3 different high-density polyethylene bottles with child resistant caps containing 35, 100, and 200 capsules each and labeled as per local regulations. Subjects were instructed to orally self- administer 2 capsules each morning in the appropriate combination for their assigned dose and treatment. Compliance was assessed by the number of study medication capsules contained in study medication containers returned at each visit.

Concomitant Medications Concomitant administration of the following medications was prohibited during the study: any urate lowering medications other than study medication, salicylates (chronic treatment with < 325 mg/day of aspirin was permitted), azathioprine, mercaptopurine, theophylline, intravenous colchicine, pyrazinamide, or sulfamethoxazole/trimethoprim. Drugs with secondary uricosuric effects such as fenofibrate, losartan, and chronic guaifenesin were permitted during

36 of 169 the trial. Patients taking concomitant colchicine were prohibited from taking macrolides or ketolides. Additionally, due to the increased risk for drug-drug interactions with colchicine and the concomitant use of P-gp or strong CYP3A4 inhibitors in patients with renal or hepatic impairment, subjects with renal or hepatic impairment were prohibited from taking colchicine during the study. Although the protocol permitted the short-term use (<4 weeks of continuous ingestion) of NSAIDs and COX-2 inhibitors, long-term (>4 continuous weeks) use of these drugs or high dose corticosteroids (>10 mg/day of prednisone or equivalent) other than protocol required prophylactic therapy supplied by the applicant was not allowed. Patients were instructed not to take any new medications (including over-the counter products) without first consulting with their investigator. All concomitant medications (including vitamin supplements, over-the-counter medications and oral herbal preparations) were to have been recorded at each visit in each subject’s case report form along with reason for use.

Gout Flare Prophylaxis and Treatment The prophylactic treatment for gout flares administered in this study was colchicine 0.6 mg QD which was to be used by all subjects for the first 6 months of study treatment. Acceptable alternative prophylactic regimens for patients unable to tolerate daily colchicine consisted of the following: colchicine 0.6 mg every other day, naproxen 250 mg twice daily with lansoprazole 15 mg QD in subjects with an estimated creatinine clearance > 50 mL/min, other NSAIDs or prednisone at the investigator’s discretion.

Patients who experienced an acute gout flare during the study were treated with an individualized anti-inflammatory regimen that included colchicine (administered via the acute flare regimen), increasing NSAID prophylactic gout therapy (e.g., 500 mg twice daily of naproxen on Day 1 of the flare followed by one additional 250 mg dose on each subsequent day until the flare subsided), or a NSAID/analgesic agent at the investigator’s discretion.

Removal of Subjects from Treatment or Assessment The protocol mandated that subjects who permanently discontinued study drug treatment were to have been contacted every 2 months for follow-up. Reasons for permanent discontinuation of study drug treatment included experiencing an adverse event that would have precluded further exposure, the occurrence of a major protocol deviation, pregnancy, lost to follow-up, voluntary withdrawal (subject wished to stop taking study medication), or other reasons that needed to be specified. A minimum of 2 documented telephone calls and 1 certified letter sent within 4 weeks of the most recently missed telephone call were to have been made prior to declaring a patient lost to follow-up in this study. To maintain study integrity, an independent search company was retained by the applicant to locate subjects lost to follow-up.

The protocol also contained the following withdrawal criteria for patients participating in this study: withdrawal of consent by the subject, the patient had a CV event that was positively adjudicated as a MACE, the required number of study MACE had occurred, or the study ended. Once a subject had experienced a CV event that was positively adjudicated as a MACE, that patient was no longer followed-up by the study site.

37 of 169

Study Procedures: The following is a tabular flow chart of the scheduled study observations and procedures:

Table 8 - Tabular Summary of Assessments and Procedures for Study 301

Source: Applicant’s Table 9.b; p.46-47 Clinical Study Report

Outcome Measures: Primary efficacy endpoint: The pre-specified primary endpoint was the time from randomization to the first occurrence of a major adverse cardiovascular event (MACE), defined as the composite of: o Cardiovascular (CV) death o Non-fatal myocardial infarction (MI)

38 of 169 o Non-fatal stroke o Unstable angina with urgent coronary revascularization

Secondary efficacy endpoints: The study protocol included several pre-specified secondary endpoints. No statistical testing hierarchy was pre-specified for secondary endpoints. This review discusses the following: • Time from randomization to the first occurrence of any Antiplatelet Trialists’ Collaborative (APTC) event, defined as the composite of CV death, nonfatal MI, and nonfatal stroke. • Time from randomization to first occurrence of each individual component of the primary MACE. • Time from randomization to the occurrence of all-cause death • Time from randomization to the occurrence of all cause death during on-treatment follow-up (from the time of randomization to the last dose of study medication)

Adjudication Methods An independent Cardiovascular Endpoints Committee (CEC) was constituted for the trial to adjudicate cardiovascular events for randomized subjects in an independent and blinded manner. The CEC was composed of 3 members with expertise in cardiovascular and cerebrovascular medicine. The CEC charter was submitted to the Agency as part of the application package for NDA 21856. This charter was amended following the May 19, 2011 submission of amendment 6 to the study protocol so that nonserious potential CV events were excluded from adjudication since nonserious AEs were not likely to be adjudicated as a MACE event nor was there enough clinical information for accurate clinical assessment of these types of events.

According to the CEC charter, the primary responsibilities of its members were to prospectively review and adjudicate in a blinded manner all deaths and serious potential cardiovascular events reported in Study 301 to determine if the reported event met the criteria for MACE. Each death and all serious CV adverse events, including any medically significant updates, were sent by the applicant or designee to the CEC for adjudication. The CEC chair was responsible for selecting CV events for full committee adjudication. Table 9 lists the MACE and non-MACE event definitions used by the CEC during their adjudication of the deaths reported in Study 301. According to the CEC charter, all deaths (etiology unknown) not attributed to the categories of CV death or to a non-CV cause, were to have been presumed CV deaths and were to be included in the CV mortality endpoint. This document also contained regulatory acceptable definitions for the non-fatal MACE events which included non-fatal myocardial infarctions, stroke, and unstable angina requiring urgent revascularization, as well as non-major adverse cardiovascular events (Non-MACE) which included cerebral revascularization, hospitalization for heart failure, arrhythmias not associated with ischemia, venous and peripheral arterial thromboembolic vents and transient ischemic attack used in the adjudication process by the CEC (information not shown).

39 of 169 Table 9: Cardiovascular Endpoint Definitions Used to Adjudicate Deaths Reported in Study 301 Cardiovascular Death 1. Sudden Cardiac Death: Death that occurs unexpectedly in a previously stable patient and includes the following deaths: • Witnessed and instantaneous without new or worsening symptoms • Witnessed within 60 minutes of the onset of new or worsening cardiac symptoms • Witnessed and attributed to an identified arrhythmia (e.g., captured on an ECG recording or witnessed on a monitor by either a medic or paramedic • Subjects unsuccessfully resuscitated from cardiac arrest or successfully resuscitated from cardiac arrest but who die within 24 hours without identification of a non-cardiac etiology • Unwitnessed death or other causes of death (information regarding the patient’s clinical status within the week preceding death should be provided) 2. Death Due to Acute Myocardial Infarction (MI): Death occurring up to 14 days after a documented acute MI (verified by either diagnostic criteria outlined for acute MI or by autopsy findings showing recent MI or recent coronary thrombus) and here there is no conclusive evidence of another cause of death. If death occurs before biochemical confirmation of myocardial necrosis can be obtained, adjudication should be based on clinical presentation and ECG evidence. Death due to MI that occurs as direct consequence of a CV investigation /procedure/operation will be classified as death due to other cardiovascular cause. 3. Death Due to Heart Failure or Cardiogenic Shock: Deaths occurring in the context of clinically worsening symptoms and/or signs of heart failure without evidence of another cause of death. This category includes sudden death during an admission for worsening hear failure. New or worsening signs and/or symptoms of CHF include any of the following: • New or increasing symptoms and/or signs of heart failure requiring the initiation of, or an increase in, treatment directed at heart failure or occurring in a patient already receiving maximal therapy for heart failure Heart failure symptoms or signs requiring continuous intravenous therapy or oxygen administration • Confinement to bed predominantly due to heart failure symptoms • Pulmonary edema sufficient to cause tachypnea and distress not occurring in the context of an acute MI or as the consequence of an arrhythmia occurring in the absence of worsening heart failure • Cardiogenic shock not occurring in the context of an acute MI or as the consequence of an arrhythmia occurring in the absence of worsening heart failure. (Cardiogenic shock is defined as SBP <90 mmHg for greater than 1 hours, not responsive to fluid resuscitation and/or heart rate correction, and felt to be secondary to cardiac dysfunction and associated with at least 1 of the following signs of hypoperfusion: cool, clammy skin or oliguria (urine output <30 ml/hr) or altered sensorium or cardiac index <2.2 L/min/m2. Cardiogenic shock can also be defined as SBP > 90 mmHg as a result of positive inotropic or vasopressor agents alone/or with mechanical support in less than 1 hour.) 4. Death due to Cerebrovascular Event (intracranial hemorrhage or non-hemorrhagic stroke): Deaths occurring up to 30 days after a suspected stroke based on clinical signs and symptoms as well as neuroimaging and/or autopsy, and where there is no conclusive evidence of another cause of death 5. Death due to Other CV Causes: Death must be fully documented CV cause not included in one of the above categories (e.g., dysrhythmia, pulmonary embolism, or other CV Intervention) Non-Cardiovascular Death: Defined as any death not covered by cardiac death or vascular death and is categorized as follows: • Pulmonary causes • Renal Causes • Gastrointestinal Causes • Infection (includes sepsis) • Non-infectious (e.g. systemic inflammatory response syndrome [SIRS]) • Malignancy (i.e., new malignancy, worsening of prior malignancy) • Hemorrhage, not intracranial Source: Cardiovascular Endpoints Committee Charter; p. 11-14.

40 of 169 Protocol Amendments

Summarized below are the 6 protocol amendments that were made to Study 301: 1. Amendment 1 (implemented on January 20, 2010) • Since all subjects were receiving ULT with the target of achieving sUA <6.0 mg/dL, secondary efficacy endpoints were not relevant in this study. In view of this, they were re-designated as additional endpoints • ECGs were removed from the safety assessments as they were being done at the screening visit and per the investigator’s discretion at the occurrence of a CV event • Projected study duration (to approximately 5 years) and number of subjects (increased to 7500 subjects) were changed to complete the study more quickly • Entry criteria were changed to permit subjects with (1) sUA > 7mg/dl or (2) sUA >6 mg/dL AND at least 1 flare in the 12 months prior to screening and/or the presence of tophi to increase participation of gout subjects in a safety study that might not have otherwise qualified due to sUA levels • Washout period for prior ULT was shortened to 7 days from 14 days to make the washout period consistent with the duration appropriate for drug clearance while avoiding a hyperuricemic state • To prevent subjects randomized to allopurinol from withdrawing due to an inadequate response to ULT, allopurinol titration (100mg increments) was added to a maximum of 400 mg in patients with moderate renal impairment and 600 mg in patients with normal/mild renal impairment. Additional visits were added at Weeks 6, 8, and 10 to permit monitoring of sUA response to patients undergoing dose titration. • To ensure patients were titrated to a dose of study treatment that achieved sUA <6.0 mg/dL for the 5-year study duration, investigators and applicant were unblinded to sUA levels until the Week 10 visit (they were blinded to study treatment) • Entry criteria for females to be > 55 years old was added and the need for FSH testing was removed (due to being postmenopausal) • Thiazide restrictions were removed since the required allopurinol titration was to be done in 100 mg increments and patients were going to be monitored every 6 months • Study Medication and Treatment Section of the protocol was revised requiring all patients to take 2 capsules of study medication daily to maintain blind in keeping with the increase in allopurinol dose and addition of placebo capsules • Addition of yearly physical exams to schedule of events • Proposed CV event rate was increased from 2.0% to 2.8% based on the limited information on true CV rates in high risk populations • Colchicine dosing recommendations for acute gout flares were updated for consistency with USPI • Removal of required lab testing of CRP and thyroid function. Hematology and urinalysis testing were to be done at screening; chemistry testing (including LFTs)

41 of 169 were to be done at screening and at least every 6 months over the course of the study • Visits at Months 9 and 15 were added for elderly patients and those with moderate renal impairment in order to monitor LFTs and renal function • Since patients were being seen at least every 6 months, mandatory telephone contacts every 2 months after the Month 12 visit for patients still receiving study medication were removed while telephone contacts every 2 months for patients who discontinued study medication and were being followed for potential CV events were to be continued. • Clarification on the collection of aspirin and NSAID use data • Clarification of the signs and symptoms data collected for documentation of acute gout flares as defined by the ACR 2. Amendment 2 (implemented on February 24, 2010) • Addition of ECGs at the Final/End of Treatment/Early Termination Visit • Entry criteria were updated requiring females to be > 55 years old and at least 2 years postmenopausal • Clarification of overdose language to reference the double-blind study medication only • Addition of a 200-count bottle of double blind study medication to ensure patients an adequate supply of study medications • Clarification that subjects who had been taking a PPI other than lansoprazole prior to study entry could continue its use 3. Amendment 3 (implemented January 26, 2011) • Clarification that patients who prematurely discontinued study drug treatment and experienced a CV event that was positively adjudicated as a MACE were no longer followed • Study sites increased from 450 to 465 sites (addition of 15 sites in Canada) • Clarification of the entry criteria to prohibit subjects entering the study more than once • Clarification of how the double-blind nature of the study was maintained in response to a FDA comment • Addition of information regarding colchicine safety and drug interactions for use by investigators in monitoring patients who were taking colchicine for prophylaxis 4. Amendment 4 (Implemented May 19, 2011) • The Health Insurance Portability and Accountability Act Authorization (HIPAA) regulation language throughout the protocol was removed as the HIPAA Privacy Rule provides federal protections for personal health information for US subjects only. 5. Amendment 5 (Implemented September 26, 2011) • Increase in number of study sites (addition of sites in Mexico) • Clarification of study drug distribution and return information for additional countries

42 of 169 • Gout Treatment Section was updated to include prescription cards in the US for treatment of gout flares past the Month 6 visit to ensure patients had access to gout flare treatment • Addition of information regarding colchicine safety and drug interactions for use by investigators in monitoring patients who were taking colchicine for prophylaxis • Clarification on dosing of the double-blind medication to ensure patients were taking it correctly 6. Amendment 6 (implemented August 27, 2013) • Projected maximum study duration was increased to 9 years due to slower than expected enrollment • Nonserious potential CV events were excluded from endpoint adjudication as nonserious AEs were not likely to be adjudicated as a MACE event nor was there enough clinical information for accurate clinical assessment (specific to the CEC charter) • Final Month 60 visit was updated to End of Study/Early Termination and the clarification of procedures for this visit was provided • Clarifications were made to Sections 7.4 and 7.5 regarding patients discontinuing study drug treatment with continued participation via follow-up contacts, patients who terminate early from both study drug and study visits, expectations for attempted contacts with patients lost to follow-up, and patients terminated early due to AEs

Amendments 1 and 2 to the study protocol were to ensure that this post-marketing required safety study satisfied the requirements outlined in the approval letter for febuxostat while incorporating the then evolving “treat-to-target” management of chronic hyperuricemia in gout patients.

Statistical Methodology for Study TMX-301 Analysis Populations The pre-specified analysis population for the cardiovascular safety endpoints was based on the Full Analysis Set (FAS), which consisted of all randomized subjects who received at least 1 dose of study medication followed from the time of randomization to the last recorded study visit, date of loss of follow-up, or event date, regardless of their treatment adherence. The statistical evaluation of the FAS followed the intention-to-treat (ITT) principle and subjects contributed to the evaluation “as randomized”. The primary and secondary analyses of all endpoints in this review were conducted based on the FAS.

The analysis of all-cause death was repeated in an “on-treatment” population, which consisted of all randomized subjects who received at least 1 dose of study medication followed from the time of randomization to the earliest of the following events: date of loss of follow-up, date of event (death), or date of last dose of study medication.

43 of 169 Hypothesis Testing The final statistical analysis plan (version: Amendment 3) was dated 02/22/2017. The proposed statistical methods were generally agreed upon by the Agency. Study TMX-301 was designed to rule out a hazard ratio risk margin for MACE (febuxostat vs. allopurinol) greater than 1.3 at one- sided 2.5% alpha level. The corresponding statistical test can be written as:

H0: HR febuxostat/allopurinol ≥ 1.3; vs. Ha: HR febuxostat/allopurinol < 1.3

Three interim analyses were pre-specified and conducted after approximately 25%, 50%, and 75% of a maximum of 624 MACE had occurred, followed by a final analysis including all 624 events. At each interim analysis, if the upper bound of the confidence interval of the hazard ratio (febuxostat vs. allopurinol) was ≤ 1.3, the non-inferiority of febuxostat relative to allopurinol was to be declared successfully met and the trial was to be stopped. The critical values for these upper bounds were based on the Lan-DeMets-O’Brien-Fleming alpha spending function. The spending function was reviewed by Dr. Abraham (see IND 58229 dated March 29, 2011 in DARRTS).

In October 2016, the Data Monitor Committee (DMC) informed the applicant that the 75% interim analysis conducted in April 2016 had met the non-inferiority criterion for the primary endpoint. An imbalance in all-cause death and CV death had been observed between the two treatment arms. The DMC recommended that the applicant continue the trial without modification until the final 624 unique events were achieved. All DMC meeting minutes were submitted to the Agency as part of the application package.

The data submitted to the Agency in support of this application were based on the final results of the trial, which included all recorded follow-up time and all observed MACE (656 events). The data of the 75% interim analysis have not been submitted to the Agency and are not discussed further in this review. Because the DMC recommendation to continue the trial was not based on results for the primary endpoint MACE, this statistical review is based on the final analysis of the trial and all confidence intervals are presented at the nominal 95% level.

Sample Size Calculation Sample size calculations were based on the following assumptions: 2.8% annual MACE event rate, 10% dropout rate, 5 years of maximum length of participation, 2.5 years of accrual time, constant proportional hazards and exponential survival curves, one-sided 2.5% alpha level, and a true hazard ratio of 1. Based on these assumptions, a total of 7500 randomized subjects (3750 in each treatment group) were needed to achieve 624 subjects with primary MACE outcome, giving 90% power to reject a hazard ratio risk margin of 1.3 with 3 interim analyses at approximately 25%, 50% and 75% of the events using the Lan-DeMets-O’Brien-Fleming alpha spending function.

Primary Analysis of MACE The primary analysis of MACE was pre-specified as a time to first event analysis. A Cox proportional hazards model stratified by baseline renal function status (normal/mild impaired

44 of 169 or moderate impaired) was conducted to evaluate the CV safety between febuxostat and allopurinol. This analysis was conducted to rule out a hazard ratio risk margin (febuxostat vs. allopurinol) greater than 1.3 at one-sided 2.5% alpha level.

The proportional hazards assumption of the primary Cox model was evaluated graphically by plotting the scaled-Schoenfeld residuals of the model against time.

Reviewer’s comments: This review is based on the final analysis of the trial. All confidence intervals are presented at the nominal two-sided 95% level. No multiplicity adjustments were made for multiple endpoints and subgroups. As discussed previously, the DMC informed the applicant that the 75% interim analysis met the non-inferiority criterion for the primary endpoint but recommended that the trial continued until 624 MACE had been observed. Data from the 75% interim analysis have not been submitted to the Agency.

Secondary Analyses of Cardiovascular Endpoints

A similar stratified Cox proportional hazards model as the primary analysis of MACE was conducted for the following secondary analyses of cardiovascular safety in the FAS unless note otherwise: o Analysis of APTC o Analysis of individual components of the primary MACE o Analysis of all-cause death o Analysis of all-cause death during on-treatment follow-up

Sensitivity Analyses of Cardiovascular Endpoints

The statistical reviewer conducted the following sensitivity analyses to further understand the CV safety associated with febuxostat:

1. Analyses of MACE, CV death, and all-cause death by various windows of time after treatment discontinuation

2. Analyses by the final titrated treatment dose and baseline renal function

These analyses were conducted to evaluate the effect of dose within each of the renal function categories on MACE, CV death, and all-cause death. Since these analyses were based on a post-randomization characteristic “final titrated dose”, the analyses results should be considered exploratory only.

45 of 169 Study Results

Patient Disposition

Study TMX-301 was conducted at 320 sites in 3 countries: United States (275), Canada (12), and Mexico (33). The trial was initiated on 4/23/2010 (first subject signed informed consent) and was completed on 7/18/2017 (last subject last contact). Figure 2 describes subject disposition for the ITT population (FAS). A total of 9994 subjects were screened, and 6198 subjects were randomized (3101 subjects to febuxostat and 3097 subjects to allopurinol). Of the 6198 subjects randomized, 8 subjects (3 with febuxostat and 5 with allopurinol) did not receive treatment. Of the 6190 subjects randomized and treated, 2689 (43%) subjects (1324 with febuxostat; 1365 with allopurinol) completed the study treatment. A subject was considered to have completed study treatment if he completed study medication until the end of the trial. The most common reasons for study treatment discontinuation were “voluntary withdrawal” (22.1%) and “AE” (14.5%). The two treatment groups were comparable in terms of the reasons for study treatment discontinuation.

Per the trial protocol, all subjects were to be followed until the last planned study visit, regardless of treatment adherence. Figure 3 describes study visits discontinuation for the ITT population. Of the 6190 randomized subjects, 3409 (55%) subjects (1703 with febuxostat; 1706 with allopurinol) completed all the planned study visits. A subject was considered to have completed the trial if he was followed until the end of the planned study visits. The most common reasons for study visit discontinuation were “voluntary withdrawal” (19.1%), “Other” (11.2%), and “Lost to follow-up” (7.2%). The proportions of reasons for study visit discontinuation were similar between the two treatment arms.

46 of 169 Figure 2.Disposition of Subjects in Trial TMX-301

Source: Clinical study report, page 86/167

47 of 169

Figure 4. Study Follow-up Until Trial Discontinuation

Source: Created by the statistical reviewer

Summary statistics in Table 11 show similar distribution of treatment exposure across both arms in the FAS population. The median treatment exposure time was 2 years for both treatment arms.

Table 11. Exposure Time in Subjects Exposed to Study Treatment (FAS)

Febuxostat Allopurinol N=3098 N=3092 Years of Treatment Exposure Mean (SD) 2.4 (1.9) 2.4 (1.9) Median 2.0 2.0 Max 6.8 6.9 Source: Created by the statistical reviewer

Disposition for Subjects Who Discontinued the Trial Table 12 shows the disposition of subjects by vital status. Of the 6190 randomized subjects, 2689 (43%) subjects completed the planned study treatment phase and 3409 (55%) subjects completed all the planned study visits. Of the 3409 subjects who completed the study visits, 442 (13%) subjects died on study. Due to the high trial discontinuation rate, at the end of trial, the applicant updated the vital status information based on public databases for subjects who were lost to follow up or prematurely discontinued the trial. The vital status collection included 2352 subjects. Of the 2352 subjects, a total of 199 (8%) subjects were found to have died (89

49 of 169 with febuxostat and 110 with allopurinol). Of the 6190 randomized subjects, 87% of subjects had confirmed vital status (86% in febuxostat and 87% in allopurinol) and 13% (821) of subjects had unknown vital status (427 subjects in febuxostat and 394 subjects in allopurinol). Table 12. Disposition of Subjects by Vital Status

Febuxostat Allopurinol N=3098 N=3092 Completed Trial 1704 1706 Alive 1461 (86%) 1507 (88%) Died 243 (14%) 199 (12%) Discontinued Trial* 1394 1386 Alive 878 (63%) 882 (64%) Died 89 (6%) 110 (8%) Unknown 427 (31%) 394 (28%) *The applicant updated subjects’ vital status based on public records Source: Created by the statistical reviewer

Summary statistics in Table 13 show a similar distribution of on-study follow-up duration and treatment exposure in subjects who prematurely discontinued study treatment between the two treatment arms. The median on-study follow-up duration was 2 years and the median exposure time to treatment was 1 year for both arms. For subjects who prematurely discontinued study visits, the median follow-up duration (1.5 years) and the median treatment exposure (1 year) were similar in both treatment arms (shown in Table 14).

Table 13 Follow-up Time and Treatment Exposure in Subjects Who Prematurely Discontinued Study Treatment

Febuxostat Allopurinol

N=1774 N=1727

Years of Follow-up* Mean (SD) 2.4 (1.8) 2.4 (1.9) Median 2.1 2.0 Max 6.9 7.0 Years of Treatment Exposure Mean (SD) 1.6 (1.5) 1.5 (1.4) Median 1.0 1.0 Max 6.5 6.5 *Follow-up until trial discontinuation Source: Created by the statistical reviewer

50 of 169

Table 14 Follow-up Time and Treatment Exposure in Subjects Who Prematurely Discontinued Study Visits

Febuxostat Allopurinol

N=1395 N=1386

Years of Follow-up* Mean (SD) 2.0 (1.7) 2.0 (1.7) Median 1.5 1.5 Max 6.9 7.0 Years of Treatment Exposure Mean (SD) 1.6 (1.5) 1.5 (1.5) Median 1.0 1.0 Max 6.8 6.9 *Follow-up until trial discontinuation Source: Created by the statistical reviewer

Statistical Reviewer’s comments: The statistical reviewer conducted sensitivity analyses to further explore (1) the additional 199 deaths that were identified among subjects who did not complete the trial and (2) 821 subjects who had unknown vital status. These analyses are discussed in Section 4.2.4.

Protocol Violations/Deviations

A total of 3911 subjects incurred one or more protocol violations and deviations over the course of this study as shown in Table 15. Overall, the two treatment groups appear to be balanced regarding the occurrence of protocol violations/deviations. The highest rate of protocol violations/deviations for both treatment groups occurred in the “Procedure Not Performed per Protocol” category which should not have had an impact on the study’s final outcome.

Table 15: Summary of Protocol Violations/Deviations for Study 301

Febuxostat Allopurinol Total (N=3101) (N=3097) (N=6198) Subjects with > 1 Significant Protocol Deviation 1966 (63%) 1945 (63%) 3911 (63%) Entered Study but Did Not Satisfy Entry Criteria 95 (3%) 89 (3%) 184 (3%) Developed Withdrawal Criteria During Study but Not Withdrawn 11 (<1%) 9 (<1%) 20 (<1%) Received Wrong Treatment and/or Dose 65 (2%) 73 (2%) 138 (2%) Received Prohibited Concomitant Meds 221 (7%) 247 (8%) 468 (7%) Procedure Not Performed Per Protocol 1723 (56%) 1663 (54%) 3386 (55%) Other1 242 (8%) 252 (8%) 494 (8%) Note: A subject can have > 1 significant protocol deviations

51 of 169 1Other category: included events such as unsigned or inappropriately completed informed consent forms, illicit drug and alcohol use, failure to complete study worksheets, temporary withholding of study meds due to elevated enzymes, abnormal lab tests or results entered erroneously or not done, erroneous screening washout of ULT, compliance issues with study meds and/or prophylactic meds, replacement of study meds due to mishandling or loss, failure to return study meds, removal of childproofing cap from study meds, wrote on study medication, tampering with study medication, delay in reporting AEs, randomized outside window, developed cancer after randomization, missed visits, lack of translator at study visit, early termination, refused treatment procedures, withdrew consent, etc. Source: Applicant’s Table 15.1.12; p. 455 Clinical Study Report

Demographics, Disease Characteristics and Activity

As summarized by the following tables (Table 16, Table 18, Table 19, and Table 19), the treatment groups within Study TMX-301 were generally well balanced with respect to baseline demographics, disease characteristics and activity.

The subjects who participated in this trial were primarily Caucasian males with a mean age of 65 years. There was substantial representation of older age groups with 36% of patients between the ages of 65-75 years and 14% greater than 75 years.

Table 16: Baseline Demographics of Subjects Enrolled in Study 301

Demographic Characteristic Febuxostat Allopurinol Total (N=3098) (N=3092) (N=6190) Age (years) Mean (SD) 65 (9) 65 (8) 65 (9) Categories (years) <65 1584 (51%) 1506 (49%) 3090 (50%) 65-75 1094 (35%) 1135 (37%) 2229(36%) >75 420 (14%) 451 (15%) 871 (14%) Gender Male 2604 (84%) 2592 (84%) 5196 (84%) Female 494 (16%) 500 (16%) 994 (16%) Race American Indian/Alaska Native 262 (9%) 234 (8%) 496 (8%) Asian 92 (3%) 96 (3%) 188 (3%) Black/African American 552 (18%) 593 (19%) 1145 (19%) Native Hawaiian/Other Pacific Islander 13 (<1%) 14 (1%) 27 (<1%) White 2160 (70%) 2140 (69%) 4300 (70%) Other 19 (1%) 15 (1%) 34 (1%) Ethnicity (Hispanic/Latino) Yes 539 (17%) 521 (17%) 1060 (17%) No 2559 (83%) 2571 (83%) 5130 (83%) Weight (Kg) Mean (SD) 101 (22) 100 (23) 100 (23) Height (cm) Mean (SD) 173 (10) 173 (10) 173 (10) Body Mass Index (BMI) [kg/m2] Mean (SD) 34 (7) 33 (7) 34 (7) Source: Applicant’s Table 15.1.4; p. 16-18 Clinical Study Report.

52 of 169 Subjects in this study also reported a high prevalence of risk factors for metabolic syndrome, hyperuricemia and cardiovascular disease which is expected since the trial’s entry criteria targeted patients with gout and cardiovascular disease. (Table 17). Overall, the treatment groups were similar with respect to the occurrence of a variety risk factors for these conditions as well as factors which increased the risk for MACE that included obesity (66%), mild/moderate renal impairment (93%), former/current history of smoking (63%) and/or consumption of alcohol (77%), use of low dose aspirin (48%), history of kidney stones (20%), and the use of NSAIDs (29%).

Table 17: Comorbid Disease Risk Factors for Subjects Enrolled in Study 301

Comorbid Disease Risk Factors Febuxostat Allopurinol Total (N=3098) (N=3092) (N=6190) Body Mass Index (BMI) [kg/m2] Mean (SD) 34 (7) 33 (7) 34 (7) <25 kg/m2 201 97%) 201 (7%) 402 (7%) 25-30 kg/m2 844 (27%) 862 (28%) 1706 (28%) >30 kg/m2 2053 (66%) 2024 (66%) 4077 (66%) Smoking History Never Smoked 1175 (38%) 1124 (36%) 2299 (37%) Ex-smoker 1533 (50%) 1553 (50%) 3086 (50%) Current smoker 390 (13%) 415 (13%) 805 (13%) Alcohol History: Never Drank 792 (26%) 784 (25%) 1576 (26%) Ex-Drinker 805 (26%) 812 (26%) 1617 (26%) Current Drinker 1501 (49%) 1496 (48%) 2997 (48%) Renal Function1 Moderately Impaired 1636 (53%) 1631 (53%) 3267 (53%) Mildly Impaired 1217 (39%) 1231 (40%) 2448 (40%) Normal 239 (8%) 228 (7%) 467 (8%) History of Kidney Stone Yes 627 (20%) 627 (20%) 1254 (20%) No 2471 (80%) 2465 (80%) 4936 (80%) Use of Low Dose Aspirin Yes 1496 (48%) 1481 (48%) 2977 (48%) No 1602 (52%) 1611 (52%) 3213 (52%) Use of Aspirin Yes 1894 (61%) 1933 (63%) 3827 (62%) No 1204 (39%) 1159 (38%) 2363 (38%) Use of NSAIDs Yes 856 (28%) 908 (29%) 1764 (29%) No 2242 (72%) 2184 (71%) 4426 (72%) Use of Clopidogrel and Other Anti-Platelet Drugs Yes 599 (19%) 627 (20%) 1226 (20%) No 2499 (81%) 2465 (80%) 4964 (80%) Renal Function: Moderately Impaired = baseline eCLcr 30-59 ml/min; Mildly impaired = eCLcr 60-89 mL/min; Normal = eCLcr >90 mL/min. There were 7 subjects with baseline eCLcr < 30 mL/min and 1 subject with missing baseline eCLcr. Source: Applicant’s Table 15.1.4; p. 16-18 Clinical Study Report.

53 of 169 The treatment groups were also generally well balanced in terms of baseline gout disease status and treatment (Table 18). The overall mean duration of disease since the first gout attack was 12 years for the study population with 60% of subjects reporting having at least 3 gout attacks in the last year. The uric acid burden was high with an overall mean sUA level of 8.71 mg/dL with 21% of the study population reporting tophi despite 62% reporting prior ULT use. The most commonly reported ULT was allopurinol (56%) which is not unexpected since this study was initiated not long after febuxostat had been initially approved for marketing in this country.

Table 18: Summary of Subjects' Gout History, Disease Status and Treatment History for Study 301

Febuxostat Allopurinol Total Disease Characteristic (N=3098) (N=3092) (N=6190) Baseline sUA Mean (SD) 8.74 (1.69) 8.69 (1.67) 8.71 (1.68) <7.0 412 (13%) 436 (14%) 848 (14%) 7.0 -<8.0 631 (20%) 620 (20%) 1251 (20%) 8.0 - <9.0 735 (24%) 759 (25%) 1494 (24%) 9.0-<10.0 666 (22%) 646 (21%) 1313 (21%) >=10.0 654 (21%) 631 (20%) 1285 (21%) Number of Years Since Gout Diagnosis Mean (SD) 12 (11) 12 (11) 12 (11) Number of Gout Flares in the Past 12 Months 1 to 3 flares 1880 (61%) 1842 (60%) 3722 (60%) 4 to 6 flares 544 (18%) 544 (18%) 1088 (18%) >6 flares 356 (12%) 409 (13%) 765 (12%) Time Since Last Gout Flare <1 Month 1017 (33%) 978 (32%) 1995 (32%) 1-<4 Months 981 (32%) 1009 (33%) 1990 (32%) 4-<6 Months 311 (10%) 353 (11%) 664 (11%) 6-<12 Months 471 (15%) 455 (15%) 926 (15%) >=1 Year 317 (10%) 296 (10%) 613 (10%) Signs and Symptoms of Previous Gout Flares Redness 2761 (89%) 2747 (89%) 5508 (89%) Tenderness 2993 (97%) 3005 (97%) 5998 (97%) Swelling 2955 (95%) 2985 (97%) 5940 (96%) Joint Warmth 2678 (86%) 2636 (85%) 5314 (86%) Other 575 (19%) 511 (17%) 1086 (18%) Pain Scale Rating of Last Gout Flare Mean (SD) 8.1 (1.9) 8.1 (1.9) 8.1 (1.9) Pain Scale Category of Last Gout Flare 0 2 (<0.1%) 1 (<0.1%) 3 (<0.1%) 1 to 5 357 (12%) 335 (11%) 692 (11%) 6 to 10 2736 (88%) 2754 (89%) 5490 (89%) Tophi No 2430 (78%) 2442 (79%) 4872 (79%) Yes 668 (22%) 650 (21%) 1312 (21%) Number of Tophi

54 of 169 Febuxostat Allopurinol Total Disease Characteristic (N=3098) (N=3092) (N=6190) Mean (SD) 4 (10) 4 (7) 4 (9) Prior Urate Lowering Therapies No 1045 (34%) 1044 (34%) 2089 (34%) Yes 1914 (62%) 1914 (62%) 3828 (62%)

Febuxostat 134 (4%) 130 (4%) 264 (4%) Allopurinol 1738 (56%) 1742 (56%) 3480 (56%) Probenecid 37 (1%) 36 (1%) 73 (1%) Other 5 (0.2%) 6 (0.2%) 11 (0.2%) Source: Applicant’s Table 15.1.5; p. 19-21 Clinical Study Report

As noted earlier, patients who participated in this study were required to have a significant cardiovascular history. As shown in Table 19, the most common cardiovascular history reported by subjects in this study were myocardial infarction (39%), cardiac revascularization (37%), hospitalized for unstable angina (28%), and percutaneous transluminal coronary angioplasty (26%). Comorbid conditions with a high prevalence in this study population that increased the risk for metabolic syndrome, hyperuricemia and cardiovascular disease included hypertension (92%), hyperlipidemia (87%), and diabetes mellitus (55%). Overall, the two treatment groups were generally similar with respect to cardiovascular history and comorbid disease conditions.

Table 19: Summary of Subjects' Cardiovascular History in Study 301

Febuxostat Allopurinol Total Disease Characteristic (N=3098) (N=3092) (N=6190) Myocardial Infarction 1197 (39%) 1231 (40%) 2428 (39%) Hospitalized Unstable Angina 855 (28%) 869 (28%) 1724 (28%) Cardiac Revascularization 1129 (36%) 1182 (38%) 2311 (37%) Cerebral Revascularization 69 (2%) 54 (2%) 123 (2%) Stroke 460 (15%) 410 (13%) 870 (14%) Hospitalized Transient Ischemic Attack 362 (12%) 291 (9%) 653 (11%) Peripheral Vascular Disease1 412 (13%) 375 (12%) 787 (13%) Ankle Brachial Index <=0.6 207 (7%) 179 (6%) 386 (6%) Revascularization 125 (4%) 112 (4%) 237 (4%) History of Claudication 242 (8%) 226 (7%) 468 (7%) Deep Vein Thrombosis 157 (5%) 128 (4%) 285 (5%) Pulmonary Embolism 77 (3%) 59 (2%) 136 (2%) Coronary Artery Bypass Graft Procedure 706 (23%) 760 (25%) 1466 (24%) Percutaneous Transluminal Coronary Angioplasty 780 (25%) 840 (27%) 1620 (26%) Cardiac Arrhythmia 743 (24%) 779 (25%) 1522 (25%) Have a Pacemaker of Defibrillator 297 (10%) 327 (11%) 624 (10%) Congestive Heart Failure1 622 (20%) 631 (20%) 1253 (20%) Class I 47 (2%) 77 (3%) 124 (2%) Class II 86 (3%) 63 (2%) 149 (2%) Class III 23 (1%) 42 (1%) 65 (1%) Class IV 4 (0.1%) 5 (0.2%) 9 (0.1%) History of Diabetes Mellitus2 1710 (55%) 1699 (55%) 3409 (55%) Retinopathy 157 (5%) 162 (5%) 319 (5%) Neuropathy 944 (31%) 939 (30%) 1883 (30%)

55 of 169 Febuxostat Allopurinol Total Disease Characteristic (N=3098) (N=3092) (N=6190) Nephropathy 351 (11%) 371 (12%) 722 (12%) Small Vessel Vascular Disease 73 (2%) 79 (3%) 152 (3%) Cardiovascular Risk Factors Hypertension 2864 (92%) 2851 (92%) 5715 (92%) Hyperlipidemia 2678 (86%) 2702 (87%) 5380 (87%) Note: Cardiovascular histories that stopped at or before the date of informed consent. 1Subject will be counted only once if > 1 category in the list selected 2Subject will be counted only once if > 1 category in the list occurred Source: Applicant’s Table 15.1.6; p. 22 of Clinical Study Report

Treatment Compliance, Concomitant Medications, and Rescue Medication Use

The protocol specified that patients’ compliance with study medication was to have been assessed by pill counts performed on the returned study medication bottles. Overall mean compliance was high for both treatment groups (>94%) with 81% subjects achieving > 90% compliance (Table 20). Overall, compliance with study medication was comparable between the two treatment groups.

Table 20: Subjects' Compliance with Randomized Study Medication for Study 301

Febuxostat Allopurinol Total (N=3098) (N=3092) (N=6190) Overall Compliance Mean (SD) 94% (12) 95% (21) 95% (17) Min, Max 11, 441 13, 800 11, 800 Compliance Category <80% 193 (6%) 193 (6%) 386 (6%) 80-90% 420 (14%) 404 (13%) 824 (13%) =>90% 2485 (80%) 2495 (81%) 4980 (81%) Source: Applicant’s Table 15.1.9.1; p. 447-448 Clinical Study Report

The protocol for this study was notable for “treat-to-target dosing” for both febuxostat and allopurinol. Based on achieving a sUA of < 6.0 mg/dL, a higher percentage of febuxostat subjects (61%) did not require titration to a higher dose based on their initial starting dose of 40 mg once a day of the drug compared to 52% of allopurinol subjects who initially received 200 mg or 300 mg once a day of allopurinol based on their baseline renal function (Table 21). Overall, 300 mg once a day was the most commonly administered dose of allopurinol (45%) in this trial. However, 48% of the patients randomized to allopurinol underwent one or more dose adjustments of 100-mg to a protocol prespecified maximum daily dose of 600 mg of allopurinol as follows: 23% 1-dose adjustment, 21% 2-dose adjustments, and 4% 3 dose-adjustments. Based on this slow dose titration schedule, 16% of subjects who received an initial of 200 mg once a day of allopurinol based on their baseline renal function successfully underwent dose adjustments to receive a final dose of 400 mg once a day which was the maximum allowable daily dose of allopurinol for patients with renal impairment as prespecified by the study protocol.

56 of 169

Table 21: Summary of Study Drug Dose Adjustments for Study 301 (All Subjects)

Febuxostat Allopurinol Dose Titration (N=3098) (N=3092) Final Dose: 3098 (100%) 3092 (100%) Febuxostat 40 mg QD 1890 (61%) Febuxostat 80 mg QD 1208 (39%)

Allopurinol 200 mg QD 674 (22%) Allopurinol 300 mg QD 1380 (45%) Allopurinol 400 mg QD 778 (25%) Allopurinol 500 mg QD 132 (4%) Allopurinol 600 mg QD 128 (4%) No Dose Adjustment: 1891 (61%) 1607 (52%) Febuxostat 40 mg QD 1890 (61%) Febuxostat 80 mg QD 1 (<0.1)*

Allopurinol 200 mg QD 674 (22%) Allopurinol 300 mg QD 933 (30%) At Least 1 Dose Adjustment: 1207 (39%) 1485 (48%) One Dose Adjustment: 1207 (39%) 723 (23%) Febuxostat 40→80 mg QD 1207 (39%)

Allopurinol 200→300 mg QD 447 (15%) Allopurinol 300 →400 mg QD 276 (9%) Two Dose Adjustments: 634 (21%) Allopurinol 200 →300→400mg QD 502 (16%) Allopurinol 300→400→500 mg QD 132 (4%) Three Dose Adjustments: 128 (4%) Allopurinol 300→400→500→600 mg QD 128 (4%) *Subject received febuxostat 80 mg at first visit (unscheduled). Source: Applicant’s Table 15.1.10; p. 450-453 of Clinical Study Report

The protocol also required patients to take prophylactic gout mediations to prevent gout flares. This information is summarized in Table 22. The majority of subjects in this study took colchicine (84%) compared to naproxen with a proton pump inhibitor (PPI) (11%). Approximately 2% of patients required at least 1 dose adjustment or change in their prophylactic gout medication over the course of the study. Overall, the usage of prophylactic gout medications during this trial appears generally similar for the two treatment groups.

Table 22: Prophylactic Gout Medication Use and Dose Adjustments for Study 301

Febuxostat Allopurinol Total Gout medication (N=3098) (N=3092) (N=6190) Initial Prophylactic Medication Colchicine 0.6 mg QD 2604 (84%) 2591 (84%) 5195 (84%) Naproxen 250 mg BID +PPI 338 (11%) 324 (11%) 662 (11%) Other* 6 (<1%) 7 (<1%) 13 (<1%) Total Number of Days of Prophylactic Dosing

57 of 169 Febuxostat Allopurinol Total Gout medication (N=3098) (N=3092) (N=6190) Mean (SD) 174 (54) 173 (54) 173 (54) Dose Adjustments from Initial Prophylactic Dosing No Dose Adjustment 2877 (93%) 2859 (93%) 5736 (93%)

Colchicine 0.6 mg QD 2544 (82%) 2542 (82%) 5086 (82%) Naproxen 250 mg BID +PPI 327 (11%) 310 (10%) 637 (10%) Other* 6 (<1%) 7 (<1%) 13 (<1%) Dose Adjustments from Initial Prophylactic Dosing One Dose Adjustment 67 (2%) 58 (2%) 125 (2%)

Colchicine → Naproxen 250 mg BID + PPI 56 (2%) 44 (1%) 100 (2%) Colchicine → Other 0 0 0 Naproxen 250 mg BID + PPI → Colchicine 11 (<1%) 14 (<1%) 25 (<1%) Naproxen 250 mg BID + PPI → Other 0 0 0 Dose Adjustments from Initial Prophylactic Dosing Two Dose Adjustments 4 (<1%) 5 (<1%) 9 (<1%)

Colchicine → Naproxen 250 mg BID + PPI → Colchicine 4 (<1%) 5 (<1%) 9 (<1%) Colchicine → Naproxen 250 mg BID + PPI → Other 0 0 0 Source: Applicant’s Table 15.1.11; p. 454 Clinical Study Report

Information regarding concomitant medications used by more than 10% of the study population was also examined (Table 23). The most commonly reported concomitant classes of medications were lipid-modifying agents, drugs acting on renin-angiotensin system, stomatological preparations, beta-blockers, analgesics and drugs used in diabetes. This information is consistent with what is typically seen in patients with underlying cardiovascular disease since this disease is commonly associated with chronic disorders such as hyperlipidemia, hypertension and diabetes mellitus.

Table 23: Concomitant Medications Taken by > 10% of Subjects in Study 301

Febuxostat Allopurinol Total WHO Drug Class (N=3098) (N=3092) (N=6190) Number of Subjects Taking > 1 Concomitant Medication 3076 (99.3%) 3082 (99.7%) 6158 (99.5%) Lipid Modifying Agents 2495 (81%) 2484 (80%) 4979 (80%) Drugs Acting on Renin-Angiotensin System 2353 (76%) 2375 (77%) 4728 (76%) Stomatological Preparations 2244 (72%) 2288 (74%) 4532 (73%) Beta-Blocking Agents 1996 (64%) 2019 (65%) 4015 (65%) Analgesics 1626 (53%) 1583 (51%) 3209 (52%) Drugs Used in Diabetes 1600 (52%) 1594 (52%) 3194 (52%) Diuretics 1536 (50%) 1480 (48%) 3016 (49%) Cardiac Therapy 1360 (44%) 1419 (46%) 2779 (45%) Antithrombotic Agents 1266 (41%) 1304 (42%) 2570 (42%) Drugs for Acid-Related Disorders 1280 (41%) 1288 (42%) 2568 (42%) Vitamins 1257 (41%) 1240 (40%) 2497 (40%) Calcium Channel Blockers 1014 (33%) 1108 (36%) 2122 (34%) Antibacterials for Systemic Use 958 (31%) 977 (32%) 1935 (31%) Antigout Preparations 805 (26%) 817 (26%) 1622 (26%)

58 of 169 Febuxostat Allopurinol Total WHO Drug Class (N=3098) (N=3092) (N=6190) Antidiarrheal, Intestinal Anti-Inflammatories/Anti-Infective Agents 761 (25%) 737 (24%) 1498 (24%) Psycholeptics 684 (22%) 678 (22%) 1362 (22%) Mineral Supplements 668 (22%) 645 (21%) 1313 (21%) Corticosteroids, Dermatological Preparations 630 (20%) 643 (21%) 1273 (21%) Psychoanaleptics 649 (21%) 608 (20%) 1257 (20%) Urologicals 564 (18%) 615 (20%) 1179 (19%) Laxatives 586 (19%) 572 (19%) 1158 (19%) Drugs for Obstructive Airway Disease 573 (19%) 573 (19%) 1146 (19%) Antihypertensives 578 (19%) 551 (18%) 1129 (18%) Antianemic Preparations 493 (16%) 505 (16%) 998 (16%) Thyroid Therapy 403 (13%) 376 (12%) 779 (13%) Anti-Inflammatory and Antirheumatic Products 363 (12%) 407 (13%) 770 (12%) Antibiotics and Chemotherapeutics for Dermatological Use 388 (13%) 357 (12%) 745 (12%) Vaccines 361 (12%) 370 (12%) 731 (12%) Other Gynecologicals 362 (12%) 365 (12%) 727 (12%) Antihistamines for Systemic Use 353 (11%) 371 (12%) 724 (12%) Cough and Cold Preparations 362 (12%) 353 (11%) 715 (12%) Antiemetics and Antinauseants 345 (11%) 356 (12%) 701 (11%) Muscle Relaxants 336 (11%) 339 (11%) 675 (11%) Note: Concomitant medications were coded using WHO Drug Versions Year 2010. Subjects may have >1 medication per drug class. Source: Applicant’s Table 15.1.8.5; p. 367-447 Clinical Study Report

Analysis Results

Primary Analysis of MACE

Results of the pre-specified primary analysis of MACE based on a stratified Cox proportional hazards model are shown in Table 24. A total of 656 MACE events (335 with febuxostat and 321 with allopurinol) were observed during the trial. The estimated hazard ratio of MACE associated with febuxostat relative to allopurinol was 1.03 with a nominal 95% confidence interval of (0.89, 1.21), showing no increased risk of MACE associated with febuxostat. Based on this result, the upper bound of the 95% confidence interval for the hazard ratio successfully ruled out a hazard ratio of MACE greater than 1.3 associated with febuxostat.

Table 24 also presents the results of analyses of the components of the primary MACE endpoint. The same stratified Cox proportional hazard model was used to calculate the time to first event analyses. For cardiovascular death, 134 events were observed in the febuxostat arm and 100 events were observed in the allopurinol arm. The estimated hazard ratio and the corresponding 95% CI associated with febuxostat relative to allopurinol was 1.34 (1.03, 1.73), which showed an increased risk of CV death associated with febuxostat. For the rest of the components of MACE, the estimated hazard ratios and 95% CIs were 0.93 (0.72, 1.21) for non- fatal MI, 1.01 (0.73, 1.41) for non-fatal stroke; and 0.86 (0.59, 1.26) for unstable angina with urgent coronary revascularization, showing no evidence of increased risk associated with

59 of 169 febuxostat. These CV deaths are discussed below in the deaths subsection of the safety review under Section 4.5.4.

Table 24 Primary Analysis of MACE – Number of Events

Febuxostat Allopurinol N=3098 N=3092 Hazard Ratio* PY=8799.5 PY=8675.7 (95% CI) MACE 335 [3.8] 321 [3.7] 1.03 (0.89, 1.21) Cardiovascular death 134 [1.5] 100 [1.1] 1.34 (1.03, 1.73) Non-fatal MI 111 [1.2] 118 [1.3] 0.93 (0.72, 1.21) Non-fatal Stroke 71 [0.8] 70 [0.8] 1.01 (0.73, 1.41) Unstable Angina with Urgent Coronary Revascularization 49 [0.5] 56 [0.6] 0.86 (0.59, 1.26) *Hazard Ratio for Febuxostat vs. Allopurinol PY=person-year until first MACE or trial discontinuation [] indicates incidence rate per 100 person-years Source: Created by the statistical reviewer

Figure 5 shows the Kaplan-Meier cumulative probability of developing MACE by treatment arm. The curves were close throughout the trial and slightly separate after approximately month 54. The assumption of proportional hazards appears to hold for the primary MACE analysis (Figure 11 in Appendix shows the plot of the scaled Schoenfeld residuals).

Figure 5. Kaplan-Meier Cumulative Probability of Primary MACE by Treatment

Source: Created by the statistical reviewer

60 of 169 Figure 6 shows the Kaplan-Meier cumulative probability of experiencing CV death by treatment arm. The curves were close in the first 12 months of the trial and then separated gradually over time. Subjects in the febuxostat arm had higher observed probability of developing CV death than those in the allopurinol arm. The assumption of proportional hazards appears to hold for the CV death analysis (Figure 12 in Appendix shows the plot of the scaled Schoenfeld residuals).

Figure 6. Kaplan-Meier Plot for Cardiovascular Death

Subjects at risk

Febuxostat 3098 2815 2535 2146 1888 1626 1410 1199 992 794 593 440 230 51 0 Allopurinol 3092 2804 2510 2119 1848 1589 1397 1165 962 797 609 434 235 57 0

Source: Created by the statistical reviewer

Secondary Analyses of Cardiovascular Endpoints

The pre-specified secondary cardiovascular endpoints that were included in this review are the composite APTC, the individual components of the primary MACE, all-cause death in the FAS, and all-cause death during on-treatment. Table 25 presents the results of the pre-specified analysis for the secondary endpoints. The analyses of the individual components of the primary MACE were already discussed in the primary analysis of MACE and are not displayed again in this section.

A total of 567 APTC events (296 with febuxostat and 271 with allopurinol) were observed during the trial. The estimated hazard ratio for APTC was 1.09 with the 95% confidence interval of (0.92, 1.28), showing no evidence of increased risk associated with febuxostat. For all-cause death, a total of 243 deaths were observed in the febuxostat arm and 199 deaths were observed in the allopurinol arm. The imbalance between the two treatment arms was mainly due to CV deaths (134 with febuxostat and 100 with allopurinol). The estimated hazard ratio of

61 of 169 1.22 with an associated 95% confidence interval of (1.01, 1.47) showed an evidence of increased risk of all-cause death associated with febuxostat. The Kaplan-Meier plot in Figure 7 indicates that the curves for the two treatment arms were close in the beginning of the trial and then separated gradually over time. Subjects in the febuxostat arm had higher observed probability of developing all-cause death than those in the allopurinol arm. A graphical check (Figure 13 in Appendix) shows that the assumption of proportional hazards appears reasonable for the all-cause death analysis.

Table 25. Secondary Analysis of Cardiovascular Endpoints – Number of Events

Febuxostat Allopurinol Hazard Ratio* N=3098 N=3092 (95% CI) APTC 296 [3.3] 271 [3.1] 1.09 (0.92, 1.28) All-Cause Death 243 [2.6] 199 [2.2] 1.22 (1.01, 1.47) CV Death 134 [1.5] 100 [1.1] 1.34 (1.03, 1.73) Non-CV Death 109 [1.2] 99 [1.1] 1.10 (0.84, 1.45) All-Cause Death (on-Treatment) 36 [0.5] 27 [0.4] 1.31 (0.80, 2.16) *Hazard Ratio for Febuxostat vs. Allopurinol [] indicates incidence rate per 100 person-years Source: Created by the statistical reviewer

The time to first event analysis for all-cause death on-treatment is summarized in Table 25. A total of 63 deaths were observed on treatment (36 with febuxostat and 27 with allopurinol). The estimated hazard ratio of all-cause death on-treatment associated with febuxostat relative to allopurinol was 1.31 with the 95% confidence interval of (0.80, 2.16), which showed a similar estimated hazard ratio but not a statistically significant evidence of increased risk of all-cause death associated with febuxostat.

62 of 169

titrated dose and renal function. These analyses were conducted to further understand the CV safety of febuxostat.

Sensitivity analyses (3) and (4) evaluate the characteristics and potential impact of the 199 deaths found by the applicant through a search of public records among subjects who discontinued the study prematurely, and of the 821 subjects with unknown vital status.

1. Analyses of MACE, CV death, and all-cause death by various windows of time after treatment discontinuation

Table 27 presents the number of MACE, CV death, and all-cause death by various windows of time after treatment discontinuation. For MACE, 390 events (59%) occurred on-treatment. Of the 266 MACE that occurred after treatment discontinuation, approximately 34% (90 out of 266) were observed during the first 30 days after discontinuation.

A high proportion of CV deaths were observed after treatment discontinuation: although a total of 234 CV deaths (134 with febuxostat and 100 with allopurinol) were observed on-study, only 37 CV deaths (23 with febuxostat and 14 with allopurinol) were observed on-treatment. Of the 197 CV deaths that occurred after treatment discontinuation, approximately 47% (92 out of 197) were observed during the first 60 days after discontinuation. A similar pattern was observed for all-cause death: the majority (86%) of events occurred after treatment discontinuation. Further discussion of the types of deaths that occurred during the various windows of time can be founded under the death subsection of the safety review under Section 4.5.4 Safety Results.

Table 27. Number of Event by the Windows of Time after Treatment Discontinuation

MACE CV Death All-Cause Death Febuxostat Allopurinol Febuxostat Allopurinol Febuxostat Allopurinol On-Study 335 321 134 100 243 199 On-treatment 191 199 23 14 36 27 After Treatment (days) 1-30 51 39 39 27 56 45 31-60 16 15 14 12 27 24 61-90 9 10 8 11 16 15 91-180 24 22 23 21 37 31 181-360 5 8 4 5 22 19 361-720 15 8 11 4 26 15 721+ 24 20 12 6 23 23 Source: Created by the statistical reviewer

Table 28 presents results of time to event analyses of CV death and all-cause death based on the same stratified Cox proportional hazards model as the primary analysis by various windows

64 of 169 of time after treatment discontinuation. For CV deaths on-treatment, the estimated hazard ratio was 1.62 with the 95% confidence interval of (0.84, 3.16). The wider range of the confidence interval was due to the small number of events (37 CV deaths). The trend for more CV deaths in the febuxostat arm was consistent in analyses on-treatment + 30 days and on- treatment + 60 days. The estimated hazard ratios were similar to the estimated hazard ratio on-study, showing evidence of increased risk of CV death associated with febuxostat. Similar trends were observed for all-cause death: more deaths occurred in the febuxostat arm across all the windows of time after treatment discontinuation. The estimated hazard ratios of all- cause death “on-treatment, “on-treatment+30 days”, and “on-treatment+ 60 days” were similar to the estimated hazard ratio on-study.

Table 28. Analysis of CV Death and All-Cause Death – Number of Events

Febuxostat Allopurinol Hazard Ratio* N=3098 N=3092 (95% CI) CV Death (on-study) 134 100 1.34 (1.03, 1.73) CV Death (on-treatment) 23 14 1.62 (0.84, 3.16) CV Death (on-treatment + 30 days) 62 41 1.49 (1.01, 2.22) CV Death (on-treatment + 60 days) 76 53 1.42 (1.00, 2.01) All-Cause Death (on-study) 243 199 1.22 (1.01, 1.47) All-Cause Death (on-treatment) 36 27 1.31 (0.80, 2.16) All-Cause Death (on-treatment + 30 days) 92 72 1.26 (0.93, 1.72) All-Cause Death (on-treatment + 60 days) 119 96 1.22 (0.94, 1.60) *Hazard Ratio for Febuxostat vs. Allopurinol Source: Created by the statistical reviewer

2. Analyses by final titrated treatment dose and baseline renal function

To further evaluate the effect of dose within each of the renal function categories, exploratory analyses by the final titrated treatment dose and baseline renal function were conducted. By the study design, subjects randomized to febuxostat initially received 40 mg dose QD and could be titrated to 80 mg QD regardless of renal function status. Subjects randomized to allopurinol with normal renal function or mild renal impairment initially received 300 mg QD and could be titrated up to 600 mg QD by 100 mg increments monthly. Subjects randomized to allopurinol with moderate renal impairment initially received 200 mg QD and could be titrated up to 400 mg QD by 100 mg increments monthly.

Table 29 shows that higher rates of MACE, CV death, and all-cause death were observed in subjects with moderate renal impairment who titrated to higher doses in both treatment arms. A smaller difference between doses was observed among subjects with normal or mild renal impairment. The results of these analyses based on a post-randomization characteristic “final titrated treatment dose” should be interpreted with caution. It is unknown whether the

65 of 169 observed elevated rates by dose were due to a dose response or to the difference between subjects who titrated to the higher dose and those who did not.

Table 29. Sensitivity Analysis by Renal Function Status and Final Dose of Treatment Drug

Febuxostat Allopurinol N=3092 N=3090 % (n/N) % (n/N) 40 mg 80 mg 200 mg 300 mg 400 mg 500 mg 600 mg 9.6% 12.7% 11.6% 8.7% 12.5% 6.8% 13.3% MACE (180/1884) (154/1208) (78/673) (120/1379) (97/778) (9/132) (17/128) 8.2% 5.0% 6.8 % 2.5% 6.7% 20% Normal (13/159) (4/80) 0/0 (10/148) (1/40) (1/15) (5/25) 7.3% 11.9% 7.1% 7.6% 6.1% 11.8% Mild Impaired (55/756) (55/461) 0/4 (55/772) (18/238) (7/115) (12/102) 11.6% 14.2% 11.7% 12.0% 15.6% 50.0% Moderately Impaired (112/969) (95/667) (78/669) (55/459) (78/500) (1/2) (0/1) 3.6% 5.4% 3.1% 1.5% 2.7% 0.8% 2.3% CV Death (68/1884) (65/1208) (21/673) (21/1379) (21/778) (1/132) (3/128) 3.1% 1.3% 0.7% 2.5% 4.0% Normal (5/159) (1/80) 0/0 (1/148) (1/40) 0/15 (1/25) 2.2% 3.7% 1.4% 2.1% 0.9% 2.0% Mild Impaired (17/756) (17/461) 0/4 (11/772) (5/238) (1/115) (2/102) 4.7% 7.0% 3.1% 4.6% 7.2% Moderately Impaired (46/969) (47/667) (21/669) (21/459) (36/500) 0/2 0/1 6.5% 9.9% 6.5% 5.3% 9.0% 4.5% 4.7% All-Cause Death (122/1884) (120/1208) (44/673) (73/1379) (70/778) (6/132) (6/128) 4.4% 6.3% 2.7% 2.5% 6.7% 4.0% Normal (7/159) (5/80) 0/0 (4/148) (1/40) (1/15) (1/25) 4.0% 5.4% 3.8% 4.2% 4.3% 4.9% Mild Impaired (30/756) (25/461) 0/4 (29/772) (10/238) (5/115) (5/102) 8.8% 13.5% 6.6% 8.7% 11.8% Moderately Impaired (85/969) (90/667) (44/669) (40/459) (59/500) 0/2 0/1 *Renal function category was based on the baseline estimated creatinine clearance value; therefore, 8 subjects (6 with febuxostat and 2 with allopurinol) with missing value were excluded from this analysis. This resulted in missing one CV death event. Source: Created by the statistical reviewer

3. Analyses incorporating 199 additional deaths identified among subjects who prematurely discontinued study visits As discussed in Section 4.2.3, the applicant searched vital status information from public databases for subjects who were lost to follow up or prematurely discontinued the trial. The applicant identified 199 additional deaths among subjects who prematurely discontinued the

66 of 169 trial (89 with febuxostat and 110 with allopurinol). No information is available on whether some of these deaths can be classified as CV death.

Table 30 summarizes the timing of these 199 deaths by windows of time after treatment discontinuation. The majority of these deaths (88%) occurred later than 180 days after treatment discontinuation. Table 31 shows that the observed on-study follow-up time and treatment exposure for these 199 subjects were similar between the two treatment arms.

A similar stratified Cox proportional hazards model as the primary analysis of MACE was conducted for all-cause deaths with the updated vital status (including the additional 199 deaths). The total number of deaths including those that occurred during the study plus those verified through the post-study vital status search was 641 (332 with febuxostat and 309 with allopurinol). The estimated hazard ratio of all-cause death including the updated vital status associated with febuxostat relative to allopurinol was 1.09 with the 95% confidence interval of (0.94, 1.28).

Table 30. Additional 199 deaths by Windows of Time after Treatment Discontinuation

All Cause Death Febuxostat Allopurinol Additional deaths 89 110 On-treatment 0 1 After Treatment (days) 1-30 2 1 31-60 1 1 61-90 3 1 91-180 6 7 181-360 9 22 361-720 23 18 721+ 45 59 Source: Created by the statistical reviewer

67 of 169 Table 31. Follow-up Time and Treatment Exposure Time for Additional 199 Deaths

Febuxostat Allopurinol

N=89 N=110

Years of Follow-up* Mean (SD) 1.7 (1.5) 1.9 (1.5) Median 1.3 1.5 Max 5.9 6.1 Years of Treatment Exposure Mean (SD) 1.3 (1.3) 1.4 (1.3) Median 0.7 0.8 Max 5.5 5.9 *Follow-up until trial discontinuation Source: Created by the statistical reviewer

4. Disposition of 821 subjects with unknown vital status The applicant was unable to obtain the vital status of 821 subjects who were lost to follow up or prematurely discontinued the trial. Summary statistics in Table 32 show that the median (1.8 years) on-study follow-up times and the median (1.0 year) treatment exposure time in these subjects were similar between the two treatment arms. The proportion of these subjects who experienced MACE prior to study discontinuation was 6.1% in the febuxostat arm and 7.1% in the allopurinol arm. The total observed patient years (on-study) for these subjects were slightly higher in the febuxostat arm (779.3 years) than in the allopurinol arm (740.8 years).

Table 32. Follow-up Time in 821 Subjects with Unknown Vital Status

Febuxostat Allopurinol N=427 N=394 Experienced MACE on study 26 (6.1%) 28 (7.1%) Years of Follow-up* Mean (SD) 1.8 (1.5) 1.9 (1.6) Median 1.4 1.4 Max 6.5 6.9 Total Observed Patient Years 779.3 740.8 Years of Treatment Exposure Mean (SD) 1.4 (1.4) 1.4 (1.5) Median 1.0 1.0 Max 6.5 6.9 *Follow-up until trial discontinuation Source: Created by the statistical reviewer

68 of 169

Figure 10. Forest Plot of Hazard Ratio for CV Death by Baseline Subgroups (2)

Events/N (%) = event rate (%) Source: Created by the statistical reviewer

4.4.3. Statistical Issues and Collective Evidence

Study TMX-301 was a multi-national, randomized, double-blind, and active-controlled cardiovascular outcomes trial designed to evaluate the cardiovascular safety of febuxostat. The primary CV endpoint was time from randomization to first CEC-adjudicated MACE (cardiovascular death, non-fatal MI, non-fatal stroke, or unstable angina with urgent coronary revascularization).

This event-driven CVOT was designed to rule out the hazard ratio of MACE associated with febuxostat relative to allopurinol greater than the risk margin of 1.3. The pre-specified primary analysis used a Cox proportional hazards model stratified by baseline renal function status (normal/mild impaired or moderate impaired) to estimate the hazard ratio of MACE associated with febuxostat.

A total of 6190 subjects were included in the intent to treat population, with 3098 randomized to febuxostat and 3092 randomized to allopurinol. The two treatment arms were well-balanced at baseline regarding demographic and clinical characteristics. The trial had high discontinuation rates: 57% of subjects prematurely discontinued study treatment, and 45% prematurely discontinued study visits. However, for subjects who prematurely discontinued study treatment or study visits, their baseline characteristic, follow-up time, and exposure time were similar between the two treatment arms.

71 of 169

A total of 656 MACE were observed during the trial; 335 MACE were observed in the febuxostat arm and 321 MACE were observed in the allopurinol arm. The pre-specified stratified Cox proportional hazard model obtained an estimated hazard ratio of febuxostat vs. allopurinol of 1.03 with a 95% confidence interval of (0.89, 1.21). The upper bound of this 95% confidence interval was below the pre-specified risk margin of 1.3. However, one of the MACE components, CV death, observed an imbalance between the two treatment arms (134 with febuxostat and 100 with allopurinol). The estimated hazard ratio of CV death and the corresponding 95% CI associated with febuxostat relative to allopurinol was 1.34 (1.03, 1.73), showing an increased risk of CV death associated with febuxostat (shown in Table 33). For all- cause death, the data also showed an increased risk associated with febuxostat, HR with 95% CI: 1.22 (1.01, 1.47). The imbalance in all-cause deaths was mainly due to CV deaths. The estimated number needed to harm to observe an additional CV death associated with febuxostat in the population in this trial was 278 patient-years and 217 patient-years for all- cause death.

Subgroup analyses showed no evidence of an increased risk of MACE associated with febuxostat in subgroups defined by gender, age, race and country of randomization. However, for CV death, subgroup analyses showed that the estimated hazard ratio of CV death associated with febuxostat was higher among subgroups of baseline NSAIDS users and aspirin non-users. Analyses results are provided in Section 4.2.5.

Table 33. Primary and Secondary Analyses of Cardiovascular Endpoints

Febuxostat Allopurinol N=3098 N=3092 Hazard Ratio* PY=8799.5 PY=8675.7 (95% CI) MACE 335 [3.8] 321 [3.7] 1.03 (0.89, 1.21) Cardiovascular death 134 [1.5] 100 [1.1] 1.34 (1.03, 1.73) Non-fatal MI 111 [1.2] 118 [1.3] 0.93 (0.72, 1.21) Non-fatal Stroke 71 [0.8] 70 [0.8] 1.01 (0.73, 1.41) Unstable Angina with Urgent Coronary Revascularization 49 [0.5] 56 [0.6] 0.86 (0.59, 1.26) APTC** 296 [3.3] 271 [3.1] 1.09 (0.92, 1.28) All Cause Death 243 [2.6] 199 [2.2] 1.22 (1.01, 1.47) All Cause Death (On-Treatment) 36 [0.5] 27 [0.4] 1.31 (0.80, 2.16) *Hazard Ratio for Febuxostat vs. Allopurinol **APTC indicates a composite of CV death, non-fatal MI, and non-fatal stroke PY=person-year until first MACE or trial discontinued [] indicates incidence rate per 100 person-years Source: Created by the statistical reviewer

4.4.4. Clinical Review of Safety

This application contained double-blind safety data collected over the nearly 8-years it took to complete the phase 3, postmarketing safety study, TMX-67-301. Since the primary objective of

72 of 169 this study was a cardiovascular safety endpoint (the MACE endpoint) that was discussed in the preceding section, the focus of the following safety review is on the remaining safety data summarized in the clinical study report for TMX-301, the Integrated Summary of Safety and electronic datasets for adverse events, lab data and vital signs. These data were examined by this medical officer for any additional or new safety signals associated with the administration of febuxostat.

Review of the Safety Database Relevant characteristics of the safety population: Overall, subjects who participated in this trial had a high CV risk profile as supported by their documented history of cardiovascular disease and comorbidities (Table 19). Additionally, this study population was older with approximately half of the subjects >65 years of age, with the majority of patients having mild/moderate renal impairment and multiple risk factors for gout and cardiovascular disease. (Refer to Table 16 and Table 17 above.) This is in contrast to the phase 3 safety database originally reviewed in support of the drug’s marketing approvable in which 60% of the subjects had 2 or more cardiovascular risk factors.

Adequacy of Applicant’s Clinical Safety Assessments Issues Regarding Data Integrity and Submission Quality The data quality submitted was well-organized and adequate to perform a complete review of safety. Several information requests were sent to the applicant during the review of safety for additional analyses which were satisfactory and received in a timely manner. Categorization of Adverse Events Verbatim terms of AEs recorded in the case report forms (CRF) by investigators were coded by the applicant using MedDRA dictionary Preferred Term (PT), High-Level Term (HLT), and System Organ Class (SOC) version 20.0. The MedDRA coding of the information generated from Study 301 conducted by the applicant was generally acceptable. Additionally, the clinical lab and vital sign ranges for clinically significant abnormal results was reviewed and appeared to be appropriate.

Safety Results All safety analyses were performed on the population who received at least 1 dose of study medication. Table 34 summarizes treatment-emergent adverse events (TEAEs) that were reported in study 301. The majority of subjects who participated in this trial experienced at least 1 TEAE over the course of the study. The proportions of subjects experiencing TEAEs, a severe TEAE, a serious TEAE or a TEAE leading to discontinuation of study medication were similar for the two treatment groups. However, numerically more deaths occurred in patients randomized to treatment with febuxostat than allopurinol. These deaths will be discussed further below.

73 of 169 Table 34: Summary of Subjects by Treatment Group Who Experienced Treatment Emergent Adverse Events and Deaths During Study 301 Febuxostat Allopurinol Total (N=3098) (N=3092) (N=6190) Any Treatment Emergent Adverse Event (TEAE)1 2571 (83%) 2540 (82%) 5111 (83%) Any Severe TEAE 959 (31%) 925 (30%) 1884 (30%) Any Serious TEAE 1046 (34%) 995 (32%) 2041 (33%) Any TEAE Leading to Study Medication Discontinuation 403 (13%) 396 (13%) 799 (13%) Deaths (All Cause) 243 (8%) 199 (6%) 442 (7%) Note: A subject was counted once if the subject reported the same event more than once. Source: Tables 15.3.1.14, 15.3.1.9.1., 15.3.1.24 and Appendix 16.2.7.1

Deaths

The applicant included an analysis of the cause of death for all patients enrolled in study 301. Out of the total 442 all-cause deaths reported in this study, 243 deaths occurred in subjects in the febuxostat group versus 199 deaths in the allopurinol group (Table 26). The majority of deaths from all-causes occurred after study treatment was discontinued. Numerically more deaths occurred after treatment discontinuation in the febuxostat group versus the allopurinol group.

All deaths and serious CV AE were adjudicated by the CEC. The following discussion of deaths is based upon the cause of death as determined by the preferred term. Table 35 is a tabular listing of TEAEs leading to death by MedDRA system organ class (SOC) and preferred term. Examination of the data displayed in this table reveals Cardiac Disorders, General Disorders and Administration Site Conditions, Respiratory, Thoracic and Mediastinal Disorders, and Neoplasms Benign, Malignant, and Unspecified were the most common categories for TEAEs resulting in death in study 301. Higher rates of deaths due to Cardiac Disorders and Respiratory Thoracic and Mediastinal Disorders not in favor of the febuxostat treatment group are noted; proportions for the remaining categories are generally similar. The predominance of deaths due to Cardiac Disorders is not unexpected in view of the design and objectives of study 301. Additional examination of deaths due to Cardiac Disorders reveals numeric imbalances in deaths attributed to cardiac arrest, congestive cardiac failure, myocardial infarction, cardio- pulmonary arrest and acute myocardial infarction in the febuxostat group versus the allopurinol group with the largest difference between the two treatment groups noted for death due to congestive cardiac failure [22 cases (0.7%) versus 13 cases (0.4%)].

The higher proportion of febuxostat deaths under the Respiratory, Thoracic and Mediastinal category is due to multiple numeric imbalances in deaths attributed primarily to respiratory failure, acute respiratory failure, and respiratory arrest. Of note, there were 2 deaths due to Hepatobiliary Disorders attributed to hepatic failure (Case 3191065 due to alcoholic cirrhosis of the liver) and acute hepatic failure (Case 3027005 granulomatous hepatitis due to disseminated sarcoidosis). The current USPI for febuxostat contains a Warning and Precaution for hepatic effects based on postmarketing reports of hepatic failure that have resulted in fatalities. Overall, the types of deaths listed in this table are consistent with the risks related to the

74 of 169 underlying and concomitant medical conditions reported by subjects who participated in study 301.

Table 35: Treatment Emergent Adverse Events Leading to Death by MedDRA System Organ Class and Preferred Term by Treatment Group for Study 301

Febuxostat Allopurinol Total MedDRA System Organ Class/Preferred Term (N=3098) (N=3092) (N=6190) Deaths from All Causes 243 (7.8%) 199 (6.4%) 442 (7.1%) On Drug 36 (1.2%) 27 (0.9%) 63 (1.0%) On Drug and Up to 30-Days After Discontinuing Study Drug 92 (3.0%) 72 (2.3%) 164 (2.6%) After Last Dose of Study Drug 207 (6.7%) 172 (5.6%) 379 (6.1%) Cardiac Disorders 113 (3.6%) 83 (2.7%) 196 (3.2%) Cardiac Arrest 22 (0.7%) 18 (0.6%) 40 (0.6%) Cardiac Failure Congestive 22 (0.7%) 13 (0.4%) 35 (0.6%) Myocardial Infarction 17 (0.5%) 15 (0.3%) 32 (0.5%) Cardio-Respiratory Arrest 14 (0.4%) 8 (0.3%) 22 (0.4%) Acute Myocardial Infarction 11 (0.4%) 9 (0.3%) 20 (0.3%) Coronary Artery Disease 4 (0.1%) 4 (0.1%) 8 (0.1%) Cardiogenic Shock 5 (0.2%) 2 (<0.1%) 7 (0.1%) Arrhythmia 3 (0.1%) 2 (0.1%) 5 (0.1%) Ischemic Cardiomyopathy 3 (0.1%) 1 (<0.1%) 4 (0.1%) Cardiac Failure 3 (0.1%) 1 (<0.1%) 4 (0.1%) Hypertensive Heart Disease 3 (0.1%) 0 3 (<0.1%) Cardiac Failure Acute 3 (0.1%) 0 3 (<0.1%) Atrial Fibrillation 1 (0.1%) 0 1 (<0.1%) Myocardial ischemia 1 (<0.1%) 3 (0.1%) 4 (0.1%) Ventricular Tachycardia 1 (<0.1%) 2 (0.1%) 3 (<0.1%) Cardiopulmonary Failure 1 (<0.1%) 1 (<0.1%) 2 (0.1%) Pulseless Electrical Activity 1 (<0.1%) 1 (<0.1%) 2 (<0.1%) Ventricular Fibrillation 1 (<0.1%) 1 (<0.1%) 2 (<0.1%) Cardiovascular Disorder 1 (<0.1%) 0 1 (<0.1%) Congestive Cardiomyopathy 1 (<0.1%) 0 1 (<0.1%) Acute Coronary Syndrome 1 (<0.1%) 0 1 (<0.1%) Atrial Flutter 1 (<0.1%) 0 1 (<0.1%) Cardiac Failure Chronic 0 2 (0.1%) 2 (<0.1%) Cardiac Ventricular Thrombosis 0 1 (<0.1%) 1 (<0.1%) Arteriosclerosis Coronary Artery 0 1 (<0.1%) 1 (<0.1%) Cor Pulmonale 0 1 (<0.1%) 1 (<0.1%) Right Ventricular Failure 0 1 (<0.1%) 1 (<0.1%) Ventricular Arrhythmia 0 1 (<0.1%) 1 (<0.1%) Gastrointestinal Disorders 1 (<0.1%) 4 (0.1%) 5 (0.1%) Intestinal Obstruction 1 ((<0.1%) 0 1 (<0.1%) Vomiting 1 (<0.1%) 0 1 (<0.1%) Intestinal Ischemia 0 1 (<0.1%) 1 (<0.1%) Large Intestine Perforation 0 1 (<0.1%) 1 (<0.1%) Gastrointestinal Hemorrhage 0 1 (<0.1%) 1 (<0.1%) Upper Gastrointestinal Hemorrhage 0 1 (<0.1%) 1 (<0.1%) General Disorders and Administration Site Conditions 36 (1.2%) 35 (1.1%) 71 (1.1%) Death 31 (1.0%) 29 (0.9%) 60 (1.0%) Sudden Cardiac Death 1 (<0.1%) 2 (<0.1%) 3 (<0.1%) Sudden Death 1 (<0.1%) 1 (<0.1%) 2 (<0.1%)

75 of 169 Febuxostat Allopurinol Total MedDRA System Organ Class/Preferred Term (N=3098) (N=3092) (N=6190) Multiple Organ Dysfunction Syndrome 1 (<0.1%) 2 (<0.1%) 3 (<0.1%) Drowning 1 (<0.1%) 0 1 (<0.1%) Systemic Inflammatory Response Syndrome 1 (<0.1%) 0 1 (<0.1%) Accidental Death 0 1 (<0.1%) 1 (<0.1%) Hepatobiliary Disorders 2 (<0.1%) 1 (<0.1%) 3 (<0.1%) Hepatic Failure 1 (<0.1%) 1 (<0.1%) 2 (<0.1%) Acute Hepatic Failure 1 (<0.1%) 0 1 (<0.1%) Infections and Infestations 17 (0.5%) 17 (0.5%) 34 (0.5%) Septic Shock 6 (0.2%) 1 (<0.1%) 7 (0.1%) Pneumonia 5 (0.2%) 8 (0.3%) 13 (0.2%) Sepsis 4 (0.1%) 3 (0.1%) 7 (0.1%) Osteomyelitis 1 (<0.1%) 0 1 (<0.1%) Influenza 1 (<0.1%) 0 1 (<0.1%) Upper Respiratory Tract Infection 1 (<0.1%) 0 1 (<0.1%) Pneumonia Bacterial 0 1 (<0.1%) 1 (<0.1%) Endocarditis 0 1 (<0.1%) 1 (<0.1%) Urosepsis 0 1 (<0.1%) 1 (<0.1%) Pneumonia Staphylococcal 0 1 (<0.1%) 1 (<0.1%) Injury, Poisoning and Procedural Complications 6 (0.2%) 6 (0.2%) 12 (0.2%) Road Traffic Accident 3 (0.1%) 1 (<0.1%) 4 (0.1%) Vascular Pseudoaneurysm 1 (<0.1%) 0 1 (<0.1%) Hip Fracture 1 (<0.1%) 0 1 (<0.1%) Fall 1 (<0.1%) 0 1 (<0.1%) Multiple Injuries 1 (<0.1%) 0 1 (<0.1%) Postoperative Thoracic Procedure Complication 1 (<0.1%) 0 1 (<0.1%) Subdural Hematoma 0 1 (<0.1%) 1 (<0.1%) Post Procedural Hemorrhage 0 1 (<0.1%) 1 (<0.1%) Carbon Monoxide Poisoning 0 1 (<0.1%) 1 (<0.1%) Toxicity to Various Agents 0 1 (<0.1%) 1 (<0.1%) Head Injury 0 1 (<0.1%) Investigations 1 (<0.1%) 0 1 (<0.1%) Blood Osmolarity Decreased 1 (<0.1%) 0 1 (<0.1%) Metabolism and Nutrition Disorders 4 (0.1%) 1 (<0.1%) 5 (0.1%) Hyperkalemia 2 (0.1%) 0 2 (<0.1%) Hyponatremia 1 (<0.1%) 0 1 (<0.1%) Fluid Overload 1 (<0.1%) 0 1 (<0.1%) Lactic Acidosis 0 1 (<0.1%) 1 (<0.1%) Neoplasm Benign, Malignant, and Unspecified (Including Cysts and Polyps) 24 (0.8%) 23 (0.7%) 47 (0.8%) Pancreatic Carcinoma 4 (0.1%) 1 (<0.1%) 5 (0.1%) Lung Neoplasm Malignant 2 (0.1%) 4 (0.1%) 6 (0.1%) Colon Cancer 2 (0.1%) 0 2 (<0.1%) Lung Carcinoma Cell Type Unspecified Stage IV 2 (0.1%) 0 2 (<0.1%) Metastases to Central Nervous System 1 (<0.1%) 2 (0.1%) 3 (<0.1%) Pancreatic Carcinoma Metastatic 1 (<0.1%) 2 (0.1%) 3 (<0.1%) Myelodysplastic Syndrome 1 (<0.1%) 1 (<0.1%) 2 (<0.1%) Squamous Cell Carcinoma 1 (<0.1%) 1 (<0.1%) 2 (<0.2%) Lymphoma 1 (<0.1%) 0 1 (<0.1%) Head and Neck Cancer 1 (<0.1%) 0 1 (<0.1%) Lung Adenocarcinoma 1 (<0.1%) 0 1 (<0.1%)

76 of 169 Febuxostat Allopurinol Total MedDRA System Organ Class/Preferred Term (N=3098) (N=3092) (N=6190) Esophageal Carcinoma 1 (<0.1%) 0 1 (<0.1%) Oropharyngeal Carcinoma 1 (<0.1%) 0 1 (<0.1%) Colon Cancer Metastatic 1 (<0.1%) 0 1 (<0.1%) Diffuse Large B-Cell Lymphoma 1 (<0.1%) 0 1 (<0.1%) Glioblastoma 1 (<0.1%) 0 1 (<0.1%) Hepatocellular Carcinoma 1 (<0.1%) 0 1 (<0.1%) Bladder Cancer 0 1 (<0.1%) 1 (<0.1%) Adenocarcinoma of Colon 0 1 (<0.1%) 1 (<0.1%) Hepatic Cancer 0 1 (<0.1%) 1 (<0.1%) Squamous Cell Carcinoma of the Oral Cavity 0 1 (<0.1%) 1 (<0.1%) Chronic Myelomonocytic Leukemia 0 1 (<0.1%) 1 (<0.1%) Adenocarcinoma 0 1 (<0.1%) 1 (<0.1%) Lund Adenocarcinoma Stage IV 0 1 (<0.1%) 1 (<0.1%) Non-Small Cell Lung Cancer Metastatic 0 1 (<0.1%) 1 (<0.1%) Ductal Adenocarcinoma of Pancreas 0 1 (<0.1%) 1 (<0.1%) Renal Cancer 0 1 (<0.1%) 1 (<0.1%) Malignant Melanoma 0 1 (<0.1%) 1 (<0.1%) Nervous System Disorders 15 (0.5%) 13 (0.4%) 28 (0.5%) Cerebrovascular Accidents 6 (0.2%) 4 (0.1%) 10 (0.2%) Hypoxic Ischemic Encephalopathy 2 (0.1%) 1 (<0.1%) 3 (<0.1%) Hemorrhage Intracranial 1 (<0.1%) 2 (0.1%) 3 (<0.1%) Hemorrhagic Stroke 1 (<0.1%) 1 (<0.1%) 2 (<0.1%) Cerebral Infarction 1 (<0.1%) 0 1 (<0.1%) Syncope 1 (<0.1%) 0 1 (<0.1%) Encephalopathy 1 (<0.1%) 0 1 (<0.1%) Hepatic Encephalopathy 1 (<0.1%) 0 1 (<0.1%) Amyotrophic Lateral Sclerosis 1 (<0.1%) 0 1 (<0.1%) Cerebral Hemorrhage 0 1 (<0.1%) 1 (<0.1%) Embolic Stroke 0 1 (<0.1%) 1 (<0.1%) Ischemic Stroke 0 1 (<0.1%) 1 (<0.1%) Metabolic Encephalopathy 0 1 (<0.1%) 1 (<0.1%) Brain Injury 0 1 (<0.1%) 1 (<0.1%) Product Issues 0 1 (<0.1%) 1 (<0.1%) Device Malfunction 0 1 (<0.1%) 1 (<0.1%) Psychiatric Disorders 0 4 (0.1%) 4 (0.1%) Completed Suicide 0 3 (0.1%) 3 (0.1%) Intentional Self-Injury 0 1 (<0.1%) 1 (<0.1%) Renal an Urinary Disorders 10 (0.3%) 8 (0.3%) 18 (0.3%) Acute Kidney Injury 6 (0.2%) 4 (0.1%) 10 (0.2%) Renal Failure 2 (0.1%) 4 (0.1%) 6 (0.1%) Chronic Kidney Disease 2 (0.1%) 0 2 (<0.1%) Respiratory, Thoracic and Mediastinal Disorders 37 (1.2%) 18 (0.6%) 55 (0.9%) Respiratory Failure 11 (0.4%) 9 (0.3%) 20 (0.3%) Acute Respiratory Failure 11 (0.4%) 5 (0.2%) 16 (0.3%) Respiratory Arrest 3 (0.1%) 0 3 (<0.1%) Acute Respiratory Distress Syndrome 2 (0.1%) 1 (<0.1%) 3 (<0.1%) Pulmonary Embolism 2 (0.1%) 1 (<0.1%) 3 (<0.1%) Chronic Obstructive Pulmonary Disease 2 (0.1%) 0 2 (<0.2%) Pneumonia Aspiration 1 (<0.1%) 1 (<0.1%) 2 (<0.1%) Chronic Respiratory Failure 1 (<0.1%) 1 (<0.1%) 2 (<0.1%)

77 of 169 Febuxostat Allopurinol Total MedDRA System Organ Class/Preferred Term (N=3098) (N=3092) (N=6190) Bronchiectasis 1 (<0.1%) 0 1 (<0.1%) Pulmonary Fibrosis 1 (<0.1%) 0 1 (<0.1%) Pulmonary Edema 1 (<0.1%) 0 1 (<0.1%) Aspiration 1 (<0.1%) 0 1 (<0.1%) Pulmonary Mass 1 (<0.1%) 0 1 (<0.1%) Pulmonary Hypertension 0 1 (<0.1%) 1 (<0.1%) Vascular Disorders 8 (0.3%) 4 (0.1%) 12 (0.2%) Arteriosclerosis 2 (0.1%) 4 (0.1%) 6 (0.1%) Aortic Aneurysm 1 (<0.1%) 0 1 (<0.1%) Aortic Aneurysm Rupture 1 (<0.1%) 0 1 (<0.1%) Aortic Stenosis 1 (<0.1%) 0 1 (<0.1%) Circulatory Collapse 1 (<0.1%) 0 1 (<0.1%) Hypovolemic Shock 1 (<0.1%) 0 1 (<0.1%) Shock 1 (<0.1%) 0 1 (<0.1%) 1Treatment emergent adverse events (TEAEs) are defined as adverse events whose onset occurs, severity worsens or intensity increases after receiving the study medication and up to 30 days after the last dose. 2The total number of AEs counts all qualifying TEAEs for subjects with deaths. A subject is counted once if the subject reported the same event more than once. Source: Applicant’s Table 15.3.1.9.1; Clinical Study Report; and Table 15.3.1.15, p. 1-20; Nov. 28, 2018 IR

As noted earlier in this review, 234 deaths out of the 442 deaths that occurred in study 301 were adjudicated as CV deaths by criteria listed in preceding Table 9 and are shown in Table 36. The most common causes of adjudicated CV death in study 301 were sudden cardiac death, deaths due to heart failure and stroke. There were numeric imbalances in adjudicated all-CV deaths for sudden death, death due to heart failure and death due to myocardial infarction not in favor of febuxostat and for death due to stroke and arrhythmia not in favor of allopurinol. These imbalances in types of CV deaths persisted in the time window of up to 30-days after last dose of study medication in both treatment groups. Of the adjudicated deaths that occurred while on study drug, the majority were attributed to sudden cardiac death in both treatment groups with numerically more occurring in the febuxostat group. No other patterns in deaths that occurred in this time window were noted.

78 of 169 Table 36: Causes of Adjudicated Cardiovascular Deaths Reported in Study 301 by Various Time Windows

Febuxostat Allopurinol Total CEC Adjudicated Cardiovascular Deaths (N=3098) (N=3092) (N=6190) All CV Deaths: 134 (4.3%) 100 (3.2%) 234 (3.8%) Sudden Cardiac Death 83 (2.7%) 56 (1.8%) 139 (2.2%) Death Due to Heart Failure 20 (0.6%) 13 (0.4%) 33 (0.5%) Death Due to Stroke 8 (0.3%) 11 (0.4%) 19 (0.3%) Death due to Myocardial Infarction 11 (0.4%) 6 (0.2%) 17 (0.3%) Death Due to Arrhythmia 7 (0.2%) 9 (0.3%) 16 (0.3%) Death Due to Valvular Heart Disease 3 (<0.1%) 2 (<0.1%) 5 (<0.1%) Death Due to Heart and Respiratory Failure 1 (<0.1%) 1 (,).1%) 2 (<0.1%) Death Due to CV Hemorrhage (Anticoagulation Related) 0 1 (<0.1%) 1 (<0.1%) Death Due to Peripheral Arterial Disease 0 1 (<0.1%) 1 (<0.1%) Other CV Death (Aortic Aneurysm Rupture) 1 (<0.1%) 0 1 (<0.1%) CV Deaths on Study Drug: 23 (0.7%) 14 (0.5%) 37 (0.6%) Sudden Cardiac Death 18 (0.6%) 10 (0.3%) 28 (0.5%) Death Due to Arrhythmia 2 (<0.1%) 1 (<0.1%) 3 (<0.1%) Death Due to Heart Failure 1 (<0.1%) 2 (<0.1%) 3 (<0.1%) Death Due to Myocardial Infarction 2 (<0.1%) 1 (<0.1%) 3 (<0.1%) CV Death Up to 30-Days After Last Dose: 62 (2.0%) 41 (1.3%) 103 (1.7%) Sudden Cardiac Death 38 (1.2%) 20 (0.6%) 58 (0.9%) Death Due to Heart Failure 10 (0.3%) 6 (0.2%) 16 (0.3%) Death Due to Myocardial Infarction 6 (0.2%) 4 (0.1%) 10 (0.2%) Death Due to Arrhythmia 3 (<0.1%) 5 (0.2%) 8 (0.1%) Death Due to Stroke 3 (<0.1%) 5 (0.2%) 8 (0.1%) Death Due to Valvular Heart Disease 2 (<0.1%) 0 2 (<0.1%) Death Due to Peripheral Arterial Disease 0 1 (<0.1%) 1 (<0.1%) Source: Applicant’s Table 15.3.1.11.2; p. 512-514 Clinical Study Report.

Table 37 summarizes the 208 adjudicated non-CV deaths reported in study 301 by MedDRA preferred term. The most common causes of non-CV deaths were attributed to unknown deaths, cancer and respiratory events. Numeric imbalances are observed in deaths attributed to respiratory events, shock and trauma for the febuxostat versus allopurinol group. A numeric imbalance is also observed for the allopurinol group compared to febuxostat for deaths due to gastrointestinal events. Review of the data in Table 37 did not reveal any potential patterns or new safety signals.

79 of 169 Table 37: Summary of Adjudicated Non-Cardiovascular Deaths by MedDRA Preferred Term by Treatment Group for Study 301

Febuxostat Allopurinol Total MedDRA Preferred Term (N=3098) (N=3092) (N=6190) Non- Cardiovascular Deaths 109 (3.5%) 99 (3.2%) 208 (3.4%) Unknow Death 25 (0.8%) 23 (0.7%) 48 (0.8%) Cancer 24 (0.8%) 23 (0.7%) 47 (0.8%) Respiratory 16 (0.5%) 9 (0.3%) 25 (0.4%) Cardiac1 8 (0.3%) 10 (0.3%) 18 (0.3%) Shock 7 (0.2%) 2 (<0.1%) 9 (0.2%) Infection 6 (0.2%) 6 (0.2%) 12 (0.2%) Sepsis 5 (0.2%) 6 (0.2%) 11 (0.2%) Trauma 5 (0.2%) 2 (<0.1%) 7 (0.1%) Renal 3 (0.1%) 5 (0.2%) 8 (0.1%) Hepatic 3 (0.1%) 1 (<0.1%) 4 (<0.1%) Cardio-Pulmonary 3 (0.1%) 1 (<0.1%) 4 (<0.1%) Neurologic 2 (<0.1%) 1 (<0.1%) 3 (<0.1%) Natural Death 1 (<0.1%) 0 1 (<0.1%) Surgical Complication 1 (<0.1%) 0 1 (<0.1%) Gastrointestinal 0 4 (0.1%) 4 (<0.1%) Metabolic 0 1 (<0.1%) 1 (<0.1%) Accidental 0 1 (<0.1%) 1 (<0.1%) Suicide 0 4 (0.1%) 4 (<0.1%) 1Not adjudicated as CV death. Source: Applicant’s Table 15.3.2.3; Clinical Study Report

Serious Adverse Events

Table 38 is an abridged summary of the non-fatal serious adverse events (SAEs) observed in study 301 by MedDRA system organ class and preferred term. Overall the proportions of patients with a SAE were similar between the two treatment groups. Numeric imbalances in the number of SAE were noted with higher incidences in the febuxostat group versus allopurinol in the Musculoskeletal Disorders and Vascular Disorders SOCs. Numeric imbalances were also observed for the allopurinol group versus febuxostat in the following SOCs: Infections and Infestations, Injury, Poisoning and Procedural Complications, and Respiratory, Thoracic and Mediastinal Disorders. Once again, the SOC with the highest number of SAEs for both treatment groups was Cardiac Disorders. Further examination of the SAEs for Cardiac Disorders reveals numeric imbalances between the two treatment groups particularly for congestive cardiac failure and angina pectoris not in favor of treatment with febuxostat and for coronary artery disease and myocardial infarction not in favor of the allopurinol group. A separate review of these serious cardiovascular events is being conducted by internal consultants in the Division of Cardiorenal Drug Products.

80 of 169 Table 38: Abridged Summary of Treatment-Emergent Serious Adverse Events (SAEs) by MedDRA System Organ Class and Preferred Term Occurring in >1% of Subjects by Randomized Treatment Group for Study 301

Febuxostat Allopurinol Total MedDRA System Organ Class /Preferred Term (N=3098) (N=3092) (N=6190) Number of Subjects with > 1 SAE 1046 (33.8%) 995 (32.2%) 2041 (33%) Blood and Lymphatic System Disorders 27 (0.9%) 27 (0.9%) 54 (0.9%) Anemia 17 (0.5%) 15 (0.5%) 32 (0.5%) Cardiac Disorders 460 (14.8%) 441 (14.3%) 901 (14.6%) Cardiac Failure Congestive 124 (4.0%) 109 (3.5%) 233 (3.8%) Angina Pectoris 74 (2.4%) 59 (1.9%) 133 (2.1%) Atrial Fibrillation 65 (2.1%) 64 (2.1%) 129 (2.1%) Acute Myocardial Infarction 63 (2.0%) 68 (2.2%) 131 (2.1%) Coronary Artery Disease 51 (1.6%) 58 (1.9%) 109 (1.8%) Myocardial Infarction 41 (1.3%) 51 (1.6%) 92 (1.5%) Angina Unstable 34 (1.1%) 33 (1.1%) 67 (1.1%) Ventricular Tachycardia 25 (0.8%) 25 (0.8%) 50 (0.8%) Cardiac Arrest 17 (0.5%) 11 (0.4%) 28 (0.5%) Ischemic Cardiomyopathy 12 (0.4%) 7 (0.2%) 19 (0.3%) Bradycardia 12 (0.4%) 6 (0.2%) 18 (0.3%) Atrial Flutter 10 (0.3%) 12 (0.4%) 22 (0.4%) Cardiac Failure 8 (0.3%) 9 (0.3%) 17 (0.3%) Cardiac Failure Acute 8 (0.3%) 9 (0.3%) 17 (0.3%) Myocardial Ischemia 8 (0.3%) 5 (0.2%) 13 (0.2%) Tachycardia 8 (0.3%) 5 (0.2%) 13 (0.2%) Sinus Node Dysfunction 8 (0.3%) 6 (0.2%) 14 (0.2%) Coronary Artery Occlusion 7 (0.2%) 8 (0.3%) 15 (0.2%) Acute Coronary Syndrome 6 (0.2%) 8 (0.3%) 14 (0.2%) Mitral Valve Incompetence 6 (0.2%) 0 6 (<0.1%) Cardiogenic Shock 5 (0.25) 5 (0.2%) 10 (0.2%) Supraventricular Tachycardia 5 (0.2%) 4 (0.1%) 9 (0.1%) Atrioventricular Block Complete 4 (0.1%) 5 (0.2%) 9 (0.1%) Arteriosclerosis Coronary Artery 4 (0.1%) 4 (0.1%) 8 (0.1%) Cardiomyopathy 4 (0.1%) 3 (<0.1%) 7 (0.1%) Cardiac Failure Chronic 4 (0.1%) 5 (0.2%) 9 (0.1%) Arrhythmia 4 (0.1%) 6 (0.2%) 10 (0.2%) Cardio-Respiratory Arrest 4 (0.1%) 2 (<0.1%) 6 (<0.1%) Ventricular Fibrillation 4 (0.1%) 6 (0.2%) 10 (0.2%) Coronary Artery Stenosis 2 (<0.1%) 7 (0.2%) 9 (0.1%) Aortic Valve Stenosis 1 (<0.1%) 4 (0.1%) 5 (<0.1%) Ear and Labyrinth Disorders 5 (0.2%) 4 (0.1%) 9 (0.1%) Vertigo 4 (0.1%) 3 (<0.1%) 7 (0.1%) Gastrointestinal Disorders 103 (3.3%) 101 (3.3%) 204 (3.3%) Gastrointestinal Hemorrhage 18 (0.6%) 17 (0.5%) 35 (0.6%) Small Intestinal Obstruction 9 (0.3%) 5 (0.2%) 14 (0.2%) Abdominal Pain 8 (0.3%) 11 (0.4%) 19 (0.3%) Pancreatitis 7 (0.2%) 3 (<0.1%) 10 (0.2%) Hiatus Hernia 5 (0.2%) 0 5(<0.1%) Upper Gastrointestinal Hemorrhage 5 (0.2%) 3 (<0.1%) 8 (0.1%) Pancreatitis Acute 4 (0.1%) 1 (<0.1%) 5 (<0.1%) Dysphagia 4 (0.1%) 4 (0.1%) 8 (0.1%)

81 of 169 Febuxostat Allopurinol Total MedDRA System Organ Class /Preferred Term (N=3098) (N=3092) (N=6190) Ileus 4 (0.1%) 0 4 (<0.1%) Diarrhea 3 (<0.1%) 4 (0.1%) 7 (0.1%) Rectal Hemorrhage 2 (<0.1%) 4 (0.1%) 6 (<0.1%) Nausea 2 (<0.1%) 4 (0.1%) 6 (<0.1%) Ascites 2 (<0.1%) 4 (0.1%) 6 (<0.1%) Gastroesophageal Reflux Disease 0 6 (0.2%) 6 (<0.1%) Lower Gastrointestinal Hemorrhage 0 4 (0.1%) 4 (<0.1%) General Disorders and Administration Site Conditions 89 (2.9%) 95 (3.1%) 184 (3.0%) Noncardiac Chest Pain 34 (1.1%) 32 (1.0%) 66 (1.1%) Chest Pain 25 (0.8%) 25 (0.8%) 50 (0.8%) Death 5 (0.2%) 1 (<0.1%) 6 (<0.1%) Asthenia 4 (0.1%) 8 (0.3%) 12 (0.2%) Chest Discomfort 4 (0.1%) 8 (0.3%) 12 (0.2%) Hepatobiliary Disorders 29 (0.9%) 31 (1.0%) 60 (1.0%) Cholecystitis Acute 7 (0.2%) 8 (0.3%) 15 (0.2%) Cholecystitis 6 (0.2%) 7 (0.2%) 13 (0.2%) Cholelithiasis 5 (0.2%) 9 (0.3%) 14 (0.2%) Hepatic Cirrhosis 1 (<0.1%) 4 (0.1%) 5 (0.1%) Infections and Infestations 222 (7.2%) 234 (7.6%) 456 (7.4%) Pneumonia 67 (2.2%) 88 (2.8%) 155 (2.5%) Cellulitis 30 (1.0%) 21 (0.7%) 51 (0.8%) Sepsis 24 (0.8%) 24 (0.8%) 48 (0.8%) Osteomyelitis 15 (0.5%) 9 (0.3%) 24 (0.4%) Urinary Tract Infection 14 (0.5%) 21 (0.7%) 35 (0.6%) Bronchitis 10 (0.3%) 9 (0.3%) 19 (0.3%) Urosepsis 9 (0.3%) 4 (0.1%) 13 (0.2%) Influenza 9 (0.3%) 3 (<0.1%) 12 (0.2%) Septic Shock 7 (0.2%) 10 (0.3%) 17 (0.3%) Diverticulosis 7 (0.2%) 9 (0.3%) 16 (0.3%) Gastroenteritis 7 (0.2%) 7 (0.2%) 14 (0.2%) Staphylococcal Infection 7 (0.2%) 3 (<0.1%) 10 (0.2%) Bacteremia 6 (0.2%) 4 (0.1%) 10 (0.2%) Gangrene 4 (0.1%) 2 (<0.1%) 6 (<0.1%) Diabetic Foot Infection 4 (0.1%) 2 (<0.1%) 6 (<0.1%) Pyelonephritis 0 4 (0.1%) 4 (<0.1%) Injury, Poisoning and Procedural Complications 62 (2.0%) 88 (2.8%) 150 (2.4%) Fall 7 (0.2%) 13 (0.4%) 20 (0.3%) Hip Fracture 6 (0.2%) 6 (0.2%) 12 (0.2%) Road traffic Accident 5 (0.2%) 5 (0.2%) 10 (0.2%) Subdural Hematoma 4 (0.1%) 6 (0.2%) 10 (0.2%) Rib Fracture 3 (<0.1% 4 (0.1%) 7 (0.1%) Head Injury 1 (<0.1) 5 (0.2%) 6 (<0.1%) Subarachnoid Hemorrhage 1 (<0.1) 4 (0.1%) 5 (<0.1%) Toxicity to Various Agents 0 4 (0.1%) 4 (<0.1%) Metabolism and Nutrition Disorders 70 (2.3%) 72 (2.3%) 142 (2.3%) Dehydration 22 (0.7%) 23 (0.7%) 45 (0.7%) Hypoglycemia 12 (0.4%) 10 (0.3%) 22 (0.4%) Hyperkalemia 9 (0.3%) 10 (0.3%) 19 (0.3%) Hyperglycemia 8 (0.3%) 4 (0.1%) 12 (0.2%) Diabetes Mellitus Inadequate Control 5 (0.2%) 3 (<0.1%) 8 (0.1%)

82 of 169 Febuxostat Allopurinol Total MedDRA System Organ Class /Preferred Term (N=3098) (N=3092) (N=6190) Gout 4 (0.1%) 6 (0.2%) 10 (0.2%) Hyponatremia 4 (0.1%) 3 (<0.1%) 7 (0.1%) Musculoskeletal and Connective Tissue Disorders 118 (3.8%) 95 (3.1%) 213 (3.4%) Osteoarthritis 40 (1.3%) 34 (1.1%) 74 (1.2%) Arthritis 12 (0.4%) 7 (0.2%) 19 0.3%) Back Pain 10 (0.3%) 7 (0.2%) 17 (0.3%) Musculoskeletal Chest Pain 8 (0.3%) 5 (0.2%) 13 (0.2%) Arthralgia 8 (0.3%) 4 (0.1%) 12 (0.2%) Lumbar Spinal Stenosis 6 (0.2%) 6 (0.2%) 12 (0.2%) Spinal Column Stenosis 6 (0.2%) 3 (<0.1%) 9 (0.1%) Rhabdomyolysis 5 (0.2%) 3 (<0.1%) 8 (0.1%) Cervical Spinal Stenosis 5 (0.2%) 3 (<0.1%) 8 (0.1%) Intervertebral Disc Protrusion 4 (0.1%) 5 (0.2%) 9 (0.1%) Pain in Extremity 1 (<0.1%) 4 (0.1%) 5 (<0.1%) Neoplasms Benign, Malignant and Unspecified (including Cysts and Polyps) 109 (3.5%) 95 (3.1%) 204 (3.3%) Prostate Cancer 11 (0.4%) 16 (0.5%) 27 (0.4%) Colon Cancer 9 (0.3%) 2 (<0.1) 11 (0.2%) Bladder Cancer 8 (0.3%) 3 (<0.1%) 11 (0.2%) Malignant Melanoma 4 (0.1%) 4 (0.1%) 8 (0.1%) Renal Cell Carcinoma 4 (0.1%) 2 (<0.1%) 6 (<0.1%) Diffuse Large B-Cell Lymphoma 4 (0.1%) 1 (<0.1%) 5 (<0.1%) Adenocarcinoma of Colon 4 (0.1%) 0 4 (<0.1%) Pancreatic Carcinoma 3 (<0.1) 4 (0.1%) 7 (0.1%) Lung Neoplasm Malignant 2 (<0.1) 5 (0.2%) 7 (0.1%) Nervous System Disorders 166 (5.4%) 167 (5.4%) 333 (5.4%) Cerebrovascular Accident 45 (1.5%) 44 (1.4%) 89 (1.4%) Syncope 30 (1.0%) 34 (1.1%) 64 (1.0%) Transient Ischemic Attack 25 (0.8%) 22 (0.7%) 47 (0.8%) Presyncope 10 (0.3%) 3 (<0.1%) 13 (0.2%) Seizure 7 (0.2%) 11 (0.4%) 18 (0.3%) Carotid Artery Stenosis 7 (0.2%) 10 (0.3%) 17 (0.3%) Dizziness 6 (0.2%) 4 (0.1%) 10 (0.2%) Cerebral Infarction 6 (0.2%) 1 (<0.1%) 7 (0.1%) Encephalopathy 5 (0.2%) 5 (0.2%) 10 (0.2%) Hemorrhage Intracranial 2 (<0.1%) 6 (0.2%) 8 (0.1%) Metabolic Encephalopathy 1 (<0.1%) 6 (0.2%) 7 (0.1%) Renal and Urinary Disorders 189 (6.1%) 179 (5.8%) 368 (5.9%) Acute Kidney Injury 139 (4.5%) 132 (4.3%) 271 (4.4%) Renal Failure 19 (0.6%) 19 (0.6%) 38 (0.6%) Chronic Kidney Disease 8 (0.3%) 4 (0.1%) 12 (0.2%) Renal Impairment 7 (0.2%) 3 (<0.1%) 10 (0.2%) End Stage Renal Disease 5 (0.2%) 2 (<0.1%) 7 (0.1%) Urinary Retention 4 (0.1%) 2 (<0.1%) 6 (<0.1%) Nephrolithiasis 2 (<0.1%) 8 (0.3%) 10 (0.2%) Renal Tubular Necrosis 2 (<0.1%) 5 (0.2%) 7 (0.1%) Ureterolithiasis 2 (<0.1%) 4 (0.1%) 6 (<0.1%) Respiratory, Thoracic and Mediastinal Disorders 156 (5.0%) 171 (5.5%) 327 (5.3%) Acute Respiratory Failure 34 (1.1%) 43 (1.4%) 77 (1.2%) Chronic Obstructive Pulmonary Disease 27 (0.9%) 35 (1.1%) 62 (1.0%)

83 of 169 Febuxostat Allopurinol Total MedDRA System Organ Class /Preferred Term (N=3098) (N=3092) (N=6190) Dyspnea 22 (0.7%) 21 (0.7%) 43 (0.7%) Pulmonary Embolism 20 (0.6%) 14 (0.5%) 34 (0.5%) Respiratory Failure 17 (0.5%) 22 (0.7%) 39 (0.6%) Pulmonary Hypertension 13 (0.4%) 23 (0.7%) 36 (0.6%) Pneumonia Aspiration 5 (0.2%) 5 (0.2%) 10 (0.2%) Asthma 5 (0.2%) 4 (0.1%) 9 (0.1%) Pulmonary Edema 4 (0.1%() 5 (0.2%) 9 (0.1%) Hypoxia 3 (<0.1%) 10 (0.3%) 13 (0.3%) Pleural Effusion 3 (<0.1%) 4 (0.1%) 7 (0.1%) Acute Respiratory Distress Syndrome 2 (<0.1%) 4 (0.1%) 6 (<0.1%) Epistaxis 1 (<0.1%) 4 (0.1%) 5 (<0.1%) Atelectasis 1 (<0.1%) 4 (0.1%) 5 (<0.1%) Pulmonary Fibrosis 1 (<0.1%) 4 (0.1%) 5 (<0.1%) Vascular Disorders 116 (3.7%) 90 (2.9%) 206 (3.3%) Deep Vein Thrombosis 16 (0.5%) 15 (0.5%) 31 (0.5%) Hypotension 15 (0.5%) 12 (0.4%) 27 (0.4%) Hypertension 11 (0.4%) 9 (0.3%) 20 (0.3%) Peripheral Vascular Disorder 10 (0.3%) 6 (0.3%) 16 (0.3%) Hypertensive Crisis 10 (0.3%) 3 (<0.1%) 14 (0.2%) Aortic Stenosis 9 (0.3%) 3 (<0.1%) 12 (0.2%) Peripheral Ischemia 7 (0.2%) 1 (<0.1%) 8 (0.1%) Hematoma 6 (0.2%) 3 (<0.1%) 9 (0.1%) Peripheral Arterial Occlusive Disease 6 (0.2%) 3 (<0.1%) 9 (0.1%) Malignant Hypertension 6 (0.2%) 1 (<0.1%) 7 (0.1%) Intermittent Claudication 6 (0.2%) 1 (<0.1%) 7 (0.1%) Orthostatic Hypotension 5 (0.2%) 7 (0.2%) 13 0.2%) Aortic Aneurysm 4 (0.1%) 11 (0.4%) 15 (0.2%) Arteriosclerosis 4 (0.1%) 3 (<0.1%) 7 (0.1%) Peripheral Artery Occlusion 1 (<0.1%) 4 (0.1%) 5 (<0.1%) 1Treatment emergent adverse events (TEAEs) are defined as adverse events whose onset occurs, severity worsens or intensity increases after receiving the study medication and up to 30 days after the last dose. 2The total number of AEs counts all qualifying serious TEAEs for subjects. A subject is counted once if the subject reported the same event more than once. Source: Applicant’s Table 15.3.1.30; p. 2627-2688

Dropouts and/or Discontinuations Due to Adverse Effects

An abridged summary of treatment emergent adverse events by SOC and preferred term that resulted in subjects discontinuing from Study 301 is shown in Table 39. Overall, the proportions of subjects who discontinued study medication due to an adverse event were similar across the treatment groups. Examination of the data displayed in this table reveals Cardiac Disorders, Gastrointestinal Disorders, Investigations and Renal and Urinary Disorders were the most common types of TEAEs resulting in patients withdrawing from study 301. Numeric imbalances in the number of TEAEs leading to discontinuation of study medication are noted with higher incidences reported for the febuxostat group for Gastrointestinal Disorders and Nervous System Disorders SOCs and for Skin and Subcutaneous Tissue Disorders and Musculoskeletal Connective Tissue Disorders SOC for the allopurinol group. The most common TEAEs leading to discontinuation of study medication in the febuxostat group were congestive heart failure,

84 of 169 cerebrovascular accident and acute kidney injury while in the allopurinol group more subjects discontinued study treatment due to rash, acute kidney injury and increased blood creatinine. Both drugs are known to cause renal AEs as well as hypersensitivity reactions that can present as a variety of skin manifestations including rash. No new or potential safety signals were identified, based on the data shown in Table 39.

Table 39: Treatment Emergent Adverse Events Leading to Discontinuation of Randomized Study Medication by MedDRA System Organ Class/Preferred Term Occurring in > 0.1% of Subjects by Randomized Treatment Group for Study 301

Febuxostat Allopurinol Total MedDRA System Organ Class/Preferred Term (N=3098) (N=3092) (N=6190) Number of Subjects with > 1 TEAE Leading to Study Drug Discontinuation 403 (13.0%) 396 (12.8%) 799 (12.9%) Blood and Lymphatic System Disorders 10 (0.3%) 12 (0.4%) 22 (0.4%) Thrombocytopenia 4 (0.1%) 3 (<0.1%) 10 (0.2%) Anemia 3 (<0.1%) 5 (0.2%) 7 (0.1%) Cardiac Disorders 83 (2.7%) 82 (2.7%) 165 (2.7%) Cardiac Failure Congestive 23 (0.7%) 15 (0.5%) 38 (0.6%) Myocardial Infarction 12 (0.4%) 16 (0.5%) 28 (0.5%) Acute Myocardial Infarction 8 (0.3%) 11 (0.4%) 19 (0.3%) Cardiac Arrest 8 (0.3%) 7 (0.2%) 15 (0.2%) Coronary Artery Disease 7 (0.2%) 7 (0.2%) 14(0.2%) Atrial Fibrillation 4 (0.1%) 7 (0.2%) 11 (0.2%) Gastrointestinal Disorders 61 (2.0%) 49 (1.6%) 110 (1.8%) Diarrhea 16 (0.5%) 7 (0.2%) 23 (0.4%) Nausea 12 (0.4%) 6 (0.2%) 18 (0.3%) Abdominal Pain 6 (0.2%) 6 (0.2%) 12 (0.2%) Gastroesophageal Reflux Disease 6 (0.2%) 4 (0.1%) 10 (0.2%) Constipation 4 (0.1%) 3 (<0.1%) 7 (0.1%) Dysphagia 3 (<0.1%) 4 (0.1%) 7 (0.1%) Abdominal Pain Upper 2 (<0.1%) 4 (0.1%) 6 (<0.1%) Gastrointestinal Hemorrhage 1 (<0.1%) 5 (0.2%) 6 (<0.1%) General Disorder and Administration Site Conditions 27 (0.9%) 30 (1.0%) 57 (0.9%) Fatigue 5 (0.2%) 3 (<0.1%) 8 (0.1%) Death 4 (0.1%) 1 (<0.1%) 5 (<0.1) Chest Pain 4 (0.1%) 3 (<0.1%) 7 (0.1%) Edema Peripheral 3 (<0.1%) 8 (0.3%) 11 (0.2%) Infections and Infestations 28 (0.9%) 33 (1.1%) 61 (1.0%) Pneumonia 8 (0.2%) 11 (0.4%) 19 (0.3%) Sepsis 5 (0.2%) 5 (0.2%) 10 (0.2%) Cellulitis 1 (<0.1%) 4 (0.1%) 5 (<0.1%) Investigations 55 (1.8%) 57 (1.8%) 112 (1.8%) Blood Creatinine Increased 14 (0.5%) 19 (0.6%) 33 (0.5%) Aspartate Aminotransferase Increased 8 (0.3%) 5 (0.2%) 13 (0.2%) Hepatic Enzyme Increased 8 (0.3%) 7 (0.2%) 15 (0.2%) Blood Creatinine Phosphokinase Increased 8 (0.3%) 4 (0.1%) 12 (0.2%) Alanine Aminotransferase Increased 5 (0.2%) 4 (0.1%) 9 (0.1%) Gamma-Glutamyltransferase Increased 5 (0.2%) 5 (0.2%) 10 (0.2%)

85 of 169 Febuxostat Allopurinol Total MedDRA System Organ Class/Preferred Term (N=3098) (N=3092) (N=6190) Liver Function Test Increased 4 (0.1%) 8 (0.3%) 12 (0.2%) Blood Urea Increased 3 (<0.1%) 8 (0.3%) 11 (0.2%) Blood Alkaline Phosphatase Increased 0 4 (0.1%) 4 (<0.1%) Musculoskeletal and Connective Tissue Disorders 21 (0.7%) 30 (1.0%) 51 (0.8%) Pain in extremity 5 (0.2%) 3 (0.1%) 8 (0.1%) Arthralgia 3 (<0.1%) 7 (0.2%) 10 (0.2%) Muscle Spasms 0 4 (0.1%) 4 (<0.1%) Nervous System Disorders 49 (1.6%) 41 (1.3%) 90 (1.5%) Cerebrovascular Accident 15 (0.5%) 7 (0.2%) 22 (0.4%) Headache 7 (0.2%) 6 (0.2%) 13 (0.2%) Dizziness 3 (<0.1%) 5 (0.2%) 8 (0.1%) Renal and Urinary Disorders 47 (1.5%) 49 (1.6%) 96 (1.6%) Acute Kidney Injury 21 (0.7%) 24 (0.8%) 45 (0.7%) Chronic Kidney Injury 8 (0.3%) 3 (<0.1%) 11 90.2%) Renal Impairment 8 (0.3%) 7 (0.2%) 15 (0.2%) Renal Failure 6 (0.2%) 7 (0.2%) 13 (0.2%) Respiratory, Thoracic and Mediastinal Disorders 3 (<0.1%) 4 (0.1%) 7 (0.1%) Respiratory Failure 6 (0.2%) 6 (0.2%) 12 (0.2%) Acute Respiratory Failure 5 (0.2%) 7 (0.2%) 12 (0.2%) Dyspnea 4 (0.1%) 5 (0.2%) 9 (0.1%) Skin and Subcutaneous Tissue Disorders 19 (0.6%) 39 (1.3%) 58 (0.9%) Rash 7 (0.2%) 27 (0.9%) 34 (0.5%) Vascular Disorders 19 (0.6%) 15 (0.5%) 34 (0.5%) Hypotension 4 (0.1%) 6 (0.2%) 10 (0.2%) 1Treatment emergent adverse events (TEAEs) are defined as adverse events whose onset occurs, severity worsens or intensity increases after receiving the study medication and up to 30 days after the last dose. 2The total number of AEs counts all qualifying TEAEs leading to study drug withdrawal for subjects. A subject is counted once if the subject reported the same event more than once. Source: Applicant’s table 15.3.1.32; p. 2692-2730

Significant Adverse Events

Table 40 is a summary of AEs by SOC observed during Study 301 by treatment arm that were rated as severe in nature by study investigators. Severity of AEs observed in this study were classified by study investigators using the following categories: • Mild: The event was transient and easily tolerated by the subject • Moderate: The event caused discomfort and interrupted the subject’s usual activities • Severe: The event caused considerable interference with the subject’s usual activities

The proportions of subjects who experienced severe TEAEs was similar for the two treatment groups. The most commonly reported severe TEAEs in study 301 were: Cardiac Disorders, Infections and Infestations, Nervous System Disorders, Renal and Urinary Disorders, Respiratory, Thoracic and Mediastinal Disorders and Gastrointestinal Disorders. The pattern of severe TEAEs mirrors that observed for the serious TEAEs, and with the exception of Infections and Infestations and Respiratory, Thoracic and Mediastinal Disorders, is similar to the data for premature discontinuation of study medication.

86 of 169 Table 40: Summary of Severe Treatment Emergent Adverse Events (TEAEs)1 by MedDRA System Organ Class in Study 3012

Febuxostat Allopurinol Total MedDRA System Organ Class (N=3098) (N=3092) (N=6190) Number of Subjects with > 1 Severe TEAE 959 (31.0%) 925 (29.9%) 1884 (30.4%) Blood and Lymphatic System Disorders 23 (0.7%) 32 (1.0%) 55 (0.9%) Cardiac Disorders 350 (11.3%) 324 (10.5%) 674 (10.9%) Congenital, Familial and Genetic Disorders 1 (<0.1%) 2 (<0.1%) 3 (<0.1%) Ear and Labyrinth Disorders 4 (0.1%) 2 (0.1%) 6 (<0.1%) Endocrine Disorders 1 (<0.1%) 4 (0.1%) 5 (<0.1%) Eye Disorders 12 (0.4%) 17 (0.5%) 29 (0.5%) Gastrointestinal Disorders 122 (3.9%) 119 (3.8%) 241 (3.9%) General Disorders and Administration Site Conditions 70 (2.3%) 86 (2.8%) 156 (2.5%) Hepatobiliary Disorders 24 (0.8%) 21 (0.7%) 45 (0.7%) Immune System Disorders 2 (<0.1%) 4 (0.1%) 6 (<0.1%) Infections and Infestations 208 (6.7%) 201 (6.5%) 409 (6.6%) Injury, Poisoning and Procedural Complications 91 (2.9%) 102 (3.3%) 193 (3.1%) Investigations 40 (1.3%) 38 (1.2%) 78 (1.3%) Metabolism and Nutrition Disorders 77 (2.5%) 65 (2.1%) 142 (2.3%) Neoplasms Benign, Malignant and Unspecified (Includes Cysts/Polys 81 (2.6%) 71 (2.3%) 152 (2.5%) Nervous System Disorders 139 (4.5%) 141 (4.6%) 280 (4.5%) Product Issues 5 (0.2%) 4 (0.1%) 9 (0.1%) Psychiatric Disorders 21 (0.7%) 19 (0.6%) 40 (0.6%) Renal and Urinary Disorders 127 (4.1%) 121 (3.9%) 248 (4.0%) Reproductive System and Breast Disorders 6 (0.2%) 11 (0.4%) 17 (0.3%) Respiratory, Thoracic and Mediastinal Disorders 121 (3.9%) 128 (4.1%) 249 (4.0%) Skin and Subcutaneous Disorders 22 (0.7%) 28 (0.9%) 50 (0.8%) Vascular Disorders 101 (3.3%) 84 (2.7%) 185 (3.0%) 1Treatment emergent adverse events (TEAEs) are defined as adverse events whose onset occurs, severity worsens or intensity increases after receiving the study medication and up to 30 days after the last dose. 2The total number of AEs counts all qualifying TEAEs for subjects. A subject is counted once if the subject reported the same event more than once. Subjects with > 1 AEs within a level of the MedDRA term, the most severe event is counted. Source Table 15.3.1.24; p. 2085-2527.

Treatment Emergent Adverse Events and Adverse Reactions

The majority of subjects in study 301 experienced a TEAE while participating in this trial. Overall, the proportions of patients who experienced a TEAE during this study were similar for the two treatment groups (Table 41). Infections and Infestations, Musculoskeletal and Connective Tissue Disorders, Gastrointestinal Disorders, Investigations, Nervous System Disorders and Cardiac Disorders were the most common types of TEAEs observed in this trial. No major differences in the proportions of TEAEs by SOC are identified on review of these data.

87 of 169 Table 41: Treatment Emergent Adverse Events1 (TEAEs) by MedDRA System Organ Class by Treatment Group for Study 3012

Febuxostat Allopurinol Total MedDRA System Organ Class (N=3098) (N=3092) (N=6190) Number of Subjects with > 1 TEAE 2571 (83.0%) 2540 (82.1%) 5111 (82.6%) Blood and Lymphatic System Disorders 209 (6.7%) 192 (6.2%) 401 (6.5%) Cardiac Disorders 694 (22.4%) 654 (21.2%) 1348 (21.8%) Congenital, Familial and Genetic Disorders 21 (0.7%) 18 (0.6%) 39 (0.6%) Ear and Labyrinth Disorders 134 (4.3%) 105 (3.4%) 239 (3.9%) Endocrine Disorders 91 (2.9%) 91 (2.9%) 182 (2.9%) Eye Disorders 221 (7.1%) 221 (7.1%) 442 (7.1%) Gastrointestinal Disorders 937 (30.2%) 886 (28.7%) 1823 (29.5%) General Disorders and Administration Site Conditions 543 (17.5%) 603 (19.5%) 1146 (18.5%) Hepatobiliary Disorders 100 (3.2%) 112 (3.6%) 212 (3.4%) Immune System Disorders 64 (2.1%) 68 (2.2%) 132 (2.1%) Infections and Infestations 1212 (39.1%) 1225 (39.6%) 2437 (39.4%) Injury, Poisoning and Procedural Complications 568 (18.3%) 590 (19.1%) 1158 (18.7%) Investigations 776 (25.0%) 749 (24.2%) 1525 (24.6%) Metabolism and Nutrition Disorders 671 (21.7%) 640 (20.7%) 1311 (21.2%) Musculoskeletal and Connective Tissue Disorders 1061 (34.2%) 1085 (35.1%) 2146 (34.7%) Neoplasms Benign, Malignant and Unspecified (Includes Cysts/Polys 255 (8.2%) 224 (7.2%) 479 (7.7%) Nervous System Disorders 724 (23.4%) 671 (21.7%) 1395 (22.5%) Product Issues 16 (0.5%) 5 (0.2%) 21 (0.3%) Psychiatric Disorders 261 (8.4%) 254 (8.2%) 515 (8.3%) Renal and Urinary Disorders 486 (15.7%) 455 (14.7%) 941 (15.2%) Reproductive System and Breast Disorders 187 (6.0%) 157 (5.1%) 344 (5.6%) Respiratory, Thoracic and Mediastinal Disorders 603 (19.5%) 627 (20.3%) 1230 (19.9%) Skin and Subcutaneous Disorders 439 (14.2%) 475 (15.4%) 914 (14.8%) Social Circumstances 2 (0.1%) 0 2 (<0.1%) Surgical and Medical Procedures 9 (0.3%) 8 (0.3%) 17 (0.3%) Vascular Disorders 478 (15.4%) 506 (16.4%) 984 (15.9%) 1Treatment emergent adverse events (TEAEs) are defined as adverse events whose onset occurs, severity worsens or intensity increases after receiving the study medication and up to 30 days after the last dose. 2The total number of AEs counts all qualifying TEAEs for subjects. A subject is counted once if the subject reported the same event more than once. Source: Applicant’s table 15.3.1.16; p. 1006-1007 Clinical Study Report

Table 42 lists common TEAEs by preferred term reported by 5% or more subjects in any treatment group during study 301. TEAEs most commonly reported by febuxostat patients in this trial were: diarrhea, arthralgia, upper respiratory tract infection, bronchitis, pain in the extremity, back pain, and hypertension. The most commonly reported TEAEs by allopurinol patients were: arthralgia, diarrhea, upper respiratory tract infection, hypertension, bronchitis, pain in the extremity and back pain. The high incidence of diarrhea in both treatment groups may be related to the concomitant administration of colchicine as gout prophylaxis during the first 6 months of study 301.

88 of 169 Table 42: Summary of Common Treatment Emergent Adverse Events1 by Preferred Term Occurring in >5% of Subjects by Treatment Group in Study 3012,3

Febuxostat Allopurinol Total MedDRA Preferred Term (PT) (N=3098) (N=3092) (N=6190) Diarrhea 300 (9.7%) 287 (9.3%) 587 (9.5%) Arthralgia 262 (8.5%) 313 (10.1%) 576 (9.3%) Upper Respiratory Tract Infection 243 (7.8%) 267 (8.6%) 510 (8.2%) Bronchitis 224 (7.2%) 220 (7.1%) 444 (7.2%) Pain in Extremity 223 (7.2%) 221 (7.1%) 444 (7.2%) Back Pain 221 (7.1%) 192 (6.2%) 413 (6.7%) Hypertension 218 (7.0%) 270 (8.7%) 488 (7.9%) Peripheral Edema 170 (5.5%) 184 (6.0%) 354 (5.7%) Cardiac Congestive Failure 169 (5.6%) 152 (4.9%) 321 (5.2%) Osteoarthritis 169 (5.5%) 181 (5.9%) 350 (5.7%) Constipation 159 (5.1%) 127 (4.1%) 286 (4.6%) Urinary Tract Infection 157 (5.1%) 191 (6.2%) 348 (5.6%) Nausea 156 (5.0%) 136 (4.4%) 292 (4.7%) Blood Creatinine Phosphokinase Increased 148 (4.8%) 161 (5.2%) 309 (5.0%) Pneumonia 139 (4.5%) 167 (5.4%) 306 (4.9%) 1Treatment emergent adverse events are defined as adverse events whose onset occurs, severity worsens or intensity increases after receiving the study medication and up to 30 days after the last dose. 2The total number of AEs counts all qualifying TEAEs for subjects. A subject is counted once if the subject reported the same event more than once. 3Preferred terms are sorted in descending order by febuxostat treatment. Source: Applicant’s table 15.3.1.17.2; p. 1088-1089 Clinical Study Report

Electrocardiograms (ECGs)

Serial 12-lead ECGs were performed on all patients who participated in study 301 at Day 1 and the end of study/early termination visits by study investigators who used the following categories to interpret the ECGs: within normal limits, abnormal but not clinically significant, or abnormal and clinically significant. Based on these criteria, 2.7 % of febuxostat patients versus 2.0 % of allopurinol patients had abnormal and clinically significant ECG readings at their final study visit or early termination visit. Review of the shift from baseline in ECG interpretations for study 301 did not reveal any potential safety issues.

QT

Data from a thorough QTc study (Study C02-023) which evaluated steady-state doses of up to 300 mg of febuxostat in healthy subjects that did not result in prolongation of the QTc interval was submitted and reviewed in support of the original NDA approval of febuxostat.

89 of 169 Efficacy Assessment While Study 301 was primarily a safety study, the applicant did collect information on efficacy. As noted earlier in this review, Study 301 was designed with “treat-to-target” dose titration of both xanthine oxidase inhibitors. By the time the majority of subjects had completed the dose titration process (Month 3), the proportion of patients achieving a sUA <6.0 mg/dL were similar for both treatment groups and remained similar over the course of the study through Month 72 (Table 43). Starting at Month 3, numerically higher proportions of subjects randomized to febuxostat achieved sUA <5.0 mg/dL as compared to patients in the allopurinol group that was maintained through the Month 72 visit (Table 43).

Table 43: Proportion of Subjects with Serum Uric Acid Levels <6.0 mg/dL and <5.0 mg/dL by Visit by Treatment Group in Study 301 (Full Analysis Set)

sUA <6.0 mg/dL sUA <5.0 mg/dL Visit Febuxostat Allopurinol Febuxostat Allopurinol (N=3098) (N=3092) (N=3098) (N=3092) Week 2 61% (1757/2892) 50% (1456/2899) 34% (978/2892) 19% (549/2899) Month 3 73% (1975/2701) 69% (1863/2686) 43% (1156/2701) 27% (716/2686) Month 6 72% (1823/2537) 66% (1680/2530) 44% (1113/2537) 28% (717/2530) Month 12 73% (1544/2131) 66% (1423/2152) 46% (980/2131) 31% (662/2152) Month 18 73% (1282/1775) 68% (1194/1757) 47% (839/1775) 33% 9571/1757) Month 24 73% (1159/1580) 68% (1052/1557) 46% (728/1580) 32% (498/1557) Month 36 73% (836/1140) 70% (776/1117) 50% (566/1140) 34% (378/1117) Month 48 72% (575/799) 73% (567/782) 49% (391/799) 38% (296/782) Month 60 76% (387/511) 72% (359/500) 54% (277/511) 40% (200/500) Month 72 75% (199/267) 75% (186/248) 58% (154/267) 44% (109/248) Source: Applicant’s Tables 15.2.6.1; p. 463; and Table 15.2.8.2, p. 483 Clinical Study Report

The applicant also provided analyses of the proportion of subjects who achieved sUA levels of <6.0 mg/dL and <5.0 mg/dL at each visit by treatment group and baseline renal function (normal renal function, and mild and moderate renal impairment) (data not shown). High proportions of subjects in both treatment groups with normal, mild and moderate renal impairment were able to achieve a sUA 6.0 mg/dL at Month 3 that ranged from 71-73% for febuxostat treated patients versus 67-73% for allopurinol treated patients. Despite fluctuations between visits, these high proportions were generally maintained over the course of the study by patients with normal renal function as well as mild to moderate renal impairment. The proportions of subjects who achieved sUA <5mg/dL at Month 3 and beyond were generally lower for both treatment groups for patients with normal, mild and moderate renal impairment and ranged from 38-45% for the febuxostat group versus 23-31% for the allopurinol group.

Study 301 also captured data regarding gout flares (Table 44). Overall, and at <1 year and >1 year of study treatment, the proportions of subjects who experienced gout flares requiring treatment were similar for both treatment groups. The overall flare rates were 0.68 flares per patient-year for the febuxostat group versus 0.63 flares per patient-year for the allopurinol group.

90 of 169

Table 44: Proportion of Subjects with Gout Flares Requiring Treatment During Study 301 (Full Analysis Set)

Febuxostat Allopurinol (N=3098) (N=3092) Overall 47% (1450/3098) 45% (1386/3092) <1 Year 41% (1272/3098) 38% (1184/3092) >1 Year 27% (597/2229) 27% (608/2232) Note: A subject who reports more than 1 gout flare during the same time interval will be counted only once for that time interval. All flares up to 30-days after a subject’s final dose of study drug were included in the analyses. Source: Applicant’s Table 15.2.7.3; p. 478 Clinical Study Report

The applicant also analyzed tophus resolution rates for the two treatment groups on an annual basis over five years in Study 301. A total of 1287 subjects who participated in this study had measurable tophi at baseline, out of which 650 patients were randomized to treatment with febuxostat versus 637 patients who received allopurinol. The overall cumulative tophus resolution rates were 13% in Year 1, 45% in Year 2, 57% in Year 3, 65% in Year 4 and 71% in Year 5 and were comparable on cross-group comparison.

Overall, the efficacy data collected in Study 301 is consistent with the findings in the original febuxostat development program.

Additional Postmarket Safety Data

There are no other pending FDA recommended post-marketing requirements for safety related concerns with febuxostat. Post-marketing safety data will continue to be assessed through routine pharmacovigilance.

In addition, as noted in the Background section, the applicant is conducting a prospective, randomized, controlled, open-label, cardiovascular outcomes post-marketing study (FAST: Febuxostat versus Allopurinol Streamlined Trial) in the UK and Denmark by the applicant at the request of the EMA. According to the applicant, this study has completed enrollment with a total of 6,142 subjects randomized and has accrued a total of 456 targeted MACE events as of January 2018. However, key results from the FAST study will not be available until May 2020.

The VA Administration’s ongoing, randomized, double blind, controlled, non-inferiority study (STOP GOUT) comparing the effectiveness of allopurinol versus febuxostat in approximately 950 hyperuricemic gout patients (including patients with stage 3 chronic kidney disease) will be another potential source of post-marketing safety data for febuxostat.

4.5. Summary and Conclusions

The Cardiovascular Safety of Febuxostat and Allopurinol in Patients with Gout and Cardiovascular Morbidities (CARES) study (TMX-67-301) was a multicenter, randomized, double-blind, active-controlled, non-inferiority cardiovascular outcome safety trial of 6,190 patients with gout and significant cardiovascular disease. This trial was conducted by the

91 of 169 applicant to fulfill a post-marketing requirement to determine whether the use of febuxostat is associated with a moderate increase in the risk of serious adverse cardiovascular outcomes as compared to the other marketed member of the xanthine oxidase class, allopurinol. The primary endpoint was a composite of major adverse cardiovascular events (MACE) that included CV death, nonfatal MI, nonfatal stroke and unstable angina with urgent coronary revascularization. Numerous secondary endpoints included the individual components of the MACE composite as well as death from any cause. The trial was designed to accrue 624 MACE events and have 90% power to rule out a hazard ratio risk margin for MACE (febuxostat vs. allopurinol) greater than 1.3 at one-sided 2.5% alpha level, assuming a true hazard ratio of 1.0.

As per treat-to-target guidelines for the management of chronic gout, doses of both urate lowering drugs were titrated to a maximum dose of 80 mg QD febuxostat and 600 mg QD allopurinol based on a goal of achieving a sUA level <6 mg/dL and on renal function. During the first 6 months of the study, subjects also received prophylactic gout therapy with either colchicine, an NSAID, or prednisone.

The two treatment arms were well balanced at baseline regarding demographic characteristics, disease history, and concomitant drug use. Patient attrition was high, with approximately 57% of randomized subjects discontinuing study drugs and 45% of subjects discontinuing study visits over the course of this study. Subject disposition was comparable between treatment arms with a median on-study follow-up time of 2.7 years and a median on-treatment follow-up of 2.0 years. Due to the high trial discontinuation rate, the applicant updated the vital status information based on public databases for subjects who were lost to follow up or who terminated the trial early. Of the 6,190 subjects randomized in this study, the last known vital status was unknown for 823 (13%) subjects.

A total of 656 MACE events (335 with febuxostat and 321 with allopurinol) were observed during this trial. The estimated hazard ratio of MACE associated with febuxostat relative to allopurinol was 1.03 with a nominal 95% confidence interval of (0.89, 1.21), which excluded the pre-specified risk margin of 1.3. However, there was an imbalance between the two treatment arms in CV death (134 with febuxostat and 100 with allopurinol). The estimated hazard ratio of CV death and the corresponding 95% CI associated with febuxostat relative to allopurinol was 1.34 (1.03, 1.73), showing an increased risk of CV death associated with febuxostat. For all- cause deaths, the data also showed an increased risk associated with febuxostat, HR with 95% CI: 1.22 (1.01, 1.47). The imbalance in all-cause deaths was mainly due to CV deaths. The estimated number needed to harm to observe an additional CV death associated with febuxostat in the high cardiovascular risk population evaluated in this trial was 278 patient- years, and 217 patient-years for all-cause death.

Subgroup analyses showed no evidence of an increase in risk of MACE associated with febuxostat in subgroups defined by gender, age, race, and country of randomization. However, for CV death, subgroup analyses showed that the estimated hazard ratio of CV death associated with febuxostat was higher among subgroups of baseline NSAID users and aspirin non-users. Additionally, higher rates of MACE, CV death, and all-cause death were observed in subjects

92 of 169 with moderate renal impairment who titrated to higher doses in both treatment arms. A smaller difference between doses was observed among subjects with normal or mild renal impairment. The results of these analyses based on a post-randomization characteristic “final titrated treatment dose” should be interpreted with caution. It is unknown whether the observed elevated rates by dose were due to a dose response or to the difference between subjects who required titration to the higher dose and those who did not.

Results from the analyses of gout endpoints (proportion of patients achieving serum uric acid [sUA] <6 mg/dL and <5 mg/dL, rate of gout flares and tophi resolution) following the completion of the dose titration process were generally similar between the two treatment groups, with the exception that a higher proportion of patients in the febuxostat group achieved a sUA<5 mg/dL over the course of the trial.

Review of the study’s safety database for febuxostat identified two areas of concerns: 1) a higher rate of all-cause deaths, and 2) a higher rate of CV deaths. These concerns are very similar to those identified by the original clinical team. That team noted during their review of the initial NDA submission that “febuxostat treated subjects had a higher rate of overall mortality, a higher rate of mortality due to CV causes, and a higher rate of cardiovascular thromboembolic events as compared to allopurinol or placebo patients.”

Review of the 442 deaths reported in this study revealed more all-cause deaths occurred in the febuxostat group (243 deaths) than in the allopurinol group (199 deaths). When examined by time-window, an imbalance in all-cause deaths not in favor febuxostat is noted during all three time windows (on study drug, up to 30-days post treatment, or after last dose of study treatment) with the majority of all-cause deaths occurring after the last dose of study treatment. Overall, Cardiac Disorders, General Disorders and Administration Site Conditions, Respiratory, Thoracic and Mediastinal Disorders, and Neoplasms Benign, Malignant and Unspecified were the most common system organ class categories for serious treatment emergent adverse events (TEAEs) resulting in death. The predominance in deaths due to Cardiac Disorders is not unexpected in view of the design and objectives of this trial. Additional examination of deaths due to Cardiac Disorders revealed numeric imbalances in deaths by preferred terms for cardiac arrest, congestive heart failure, myocardial infarction, cardiopulmonary arrest and acute myocardial infarction in the febuxostat group as compared to the allopurinol group. The largest difference between the two treatment groups was CV death due to congestive cardiac failure [febuxostat: 22 cases (0.7%) versus allopurinol: 13 cases (0.4%)].

Further examination of the 234 adjudicated CV deaths showed a higher number of adjudicated CV deaths occurred in subjects treated with febuxostat (134 adjudicated CV deaths) versus allopurinol (100 adjudicated CV deaths). The majority of CV deaths (84%) occurred off- treatment with only 37 CV deaths (16%) actually occurring while subjects were taking study medication. Adjudicated CV deaths in both groups were most commonly attributed to sudden death, and deaths due to heart failure and stroke and were associated with numeric imbalances that were once again not in favor of febuxostat. A consultative review of these CV data

93 of 169 including CV deaths by the Division of Cardiovascular and Renal Products (see appended consult), stated: “Overall, the baseline CV history including congestive heart failure were similar between febuxostat and allopurinol treatment arms. Review of MedDRA data grouped by cardiac SMQs as opposed to only preferred terms did not elucidate an obvious explanation for the CV death imbalance with febuxostat.”

To better understand the CV death signal associated with febuxostat in this trial, the agency’s DCRP consultant analyzed the non-serious and serious TEAE data contained in the study’s safety database by SMQ (narrow) for the following CV events: heart failure, embolic/thrombotic events, and arrhythmias. Based on the results of these analyses, the consultant noted little or no differences between the two treatment groups for these selected CV events. The DCRP consultant also reviewed the nonclinical evidence for a febuxostat-related effect associated with these cardiac TEAEs and noted the following: • Animal models studying febuxostat did not indicate any direct effect on myocardial function or hemodynamics suggestive of a harmful mechanism nor a propensity for febuxostat to affect coagulation or platelet aggregation • In vitro and in vivo studies did not show significant effects of febuxostat on sodium, potassium, or calcium even at supratherapeutic concentrations • In vitro studies provided supportive evidence that febuxostat would not have the adverse effect of Torsades de pointes • In vitro studies in dogs with febuxostat up to 100 mg/kg (15 times higher than therapeutic dose) did not show effects on cardiovascular parameters including ECG changes

Due to the high rates of subject withdrawal and discontinuation of study medication, the study’s database was also examined for any safety signals or patterns of concern related to study attrition. Overall, the proportions of subjects who discontinued study medication due to a TEAE were similar across the treatment groups, and no new safety or potential safety signals were identified based on this analysis. Cardiac disorders, gastrointestinal disorders, investigations and renal and urinary disorders were the most common categories of TEAEs resulting in patients withdrawing from the trial. The most common TEAEs leading to discontinuation of study mediation in the febuxostat group were congestive heart failure, cerebrovascular accident and acute kidney injury. In the allopurinol group, more subjects discontinued study treatment due to rash, acute kidney injury, and increased blood creatinine. Both of these XOI drugs are known to cause renal AEs as well as hypersensitivity reactions that mainly present as skin reactions.

Results from the analysis for severe TEAEs were similar to those for discontinuation of study medication. No safety signals or concerns were identified on review of the analyses for overall TEAEs and common TEAEs occurring in > 5% of subjects. Of note, the high incidence of diarrhea in both treatment groups is most likely due to the concomitant administration of colchicine as gout prophylaxis during the first six months of the trial. No clinically significant trends were noted on review of the lab test results, or analyses of serial vital signs or 12-lead electrocardiograms contained in the study’s safety database. Based on the clinical review of

94 of 169 the overall safety for study 301, no updates to the existing Warnings and Precaution statements in the febuxostat label for hepatic effects and serious skin reactions are warranted.

Overall, study 301 was well-designed to generate the CV safety data to address the outstanding post-marketing requirement to determine whether the use of febuxostat is associated with a moderate increase in the risk of serious adverse cardiovascular outcomes as compared to allopurinol. However, the results provide supportive evidence for safety concerns that were identified by the original clinical team regarding the increase in all-cause mortality and cardiovascular mortality observed in patients treated with febuxostat versus allopurinol and placebo. These safety concerns were noted during the first review cycle of the pivotal phase 3 safety and efficacy trials submitted in support of the drug’s overall risk-benefit assessment and marketing approval. Limitations on the results from study 301 include the generalizability of these results to the broader gout population, since the subjects evaluated in study 301, by design, were at higher risk for cardiovascular disease than previously studied.

Given that febuxostat is the only XOI alternative to allopurinol, the limited number of urate lowering therapies (ULT) that are available (which all have their own inherent safety risks), and the growing gout epidemic, there is an unmet medical need for safe and efficacious urate lowering therapies. The advisory committee panel will be asked to discuss the cardiovascular safety findings from TMX-67-301, and whether the risk-benefit profile of febuxostat has changed, and what risk mitigation actions (if any) they suggest the FDA should put into place to ensure the drug’s safe use.

95 of 169

4.6. Appendices

4.6.1. References

1,6Zhu Y, Pandya BJ, Choi HK, Prevalence of gout and hyperuricemia in the US general population: the National Health and Nutrition Examination Survey 2007-2008. Arthritis Rheum 2011; 63:3136-3141.

3 White WB, Saag KG, Becker MA, Cardiovascular Safety of Febuxostat or Allopurinol in Patients with Gout, N Engl J Med. 2018 Mar 29;378(13):1200-1210.

4Zineh I, Mummameni P, Lyndly J, La Grenade LA, Chang SH, et al. Allopurinol pharmacogenetics: assessment of potential clinical usefulness. Pharmacogenomics 2011:12:1741-9.

5 Bredemeier M, Lopes LM, Eisenreich MA, Hickmann S, Bongiorno GK, d’Avila R. Xanthine oxidase inhibitors for prevention of cardiovascular events: a systemic review and meta-analysis of randomized controlled trials. BMC Cardiovasc Disord. 2018; 18:24.

96 of 169 4.6.2. Assessment of Proportional Hazards

Figure 11. Assessment of Proportional Hazards Assumption: Schoenfeld Residual Plot (MACE)

Source: Created by the statistical reviewer

Figure 12. Assessment of Proportional Hazards Assumption: Schoenfeld Residual Plot (CV Death)

Source: Created by the statistical reviewer

97 of 169 Figure 13. Assessment of Proportional Hazards Assumption: Schoenfeld Residual Plot (All-Cause Death)

Source: Created by the statistical reviewer

98 of 169 Date: 04 DEC 2018

From: Shetarra Walker MD, MSCR Division of Cardiovascular and Renal Products

Through: Martin Rose MD, JD Mary Ross Southworth, PharmD Norman Stockbridge MD, PhD Division of Cardiovascular and Renal Products

To: Sally Seymour, MD Division Director Division of Pulmonary, Allergy, and Rheumatology Products

Subject: Consult Response to DPARP to Comment on Cardiovascular Outcome Study for Uloric (febuxostat) Tablets

This memo responds to your consult to us received on 23 Jul 2018 requesting that we review and comment on a cardiovascular outcomes study (CVOT) submitted under a postmarketing requirement (PMR) for new drug application (NDA) 21856, Uloric (febuxostat) tablets.

To address your consult, we reviewed the following:  Your consult request to us  Clinical overview, clinical study report, study synopsis, and summary safety data located in Submission Number 0144 (file:////CDSESUB1/evsprod/NDA021856/0144)  Antecedent consults regarding cardiovascular (CV) safety signals our division provided to DPARP on febuxostat tablets  Citizen Petition pertaining to febuxostat tablets dated 21 Jun 2018  Antecedent clinical reviews pertaining to CV safety concerns for febuxostat tablets  Antecedent internal, sponsor, and Advisory Committee meeting minutes pertaining to CV safety concerns for febuxostat  Information Request Responses (SEQN 0160 and 0161) from the Sponsor, Takeda, dated 09 Oct 2018 and 24 Oct 2018

99 of 169 1

Reference ID: 4358691 Summary and Conclusion

We evaluated CV safety information summarized in the clinical study report for study TMX-67-301, “A Multicenter, Randomized, Active-Control, Phase 3b Study to Evaluate the Cardiovascular Safety of Febuxostat and Allopurinol in Subjects with Gout and Cardiovascular Morbidities (CARES),” additional CV safety data tables provided by the sponsor from the CARES study, and CV safety data from prior Phase 2/3 clinical studies. As expected, the cardiovascular morbidity in subjects enrolled into CARES was moderate to high. The CARES study showed a statistically significant 34% increase in the rate of CV death with febuxostat compared to allopurinol. However, the increase in CV deaths with febuxostat was not associated with an increase in the rate of serious CV events. In addition, data from previous controlled studies of febuxostat do not show a consistent pattern of findings for CV death and CV events, with one large allopurinol-controlled study (F-GT06-153) showing no signal of CV harm for febuxostat for either deaths or serious CV events. Amongst our clinical review team, the opinions vary on how to interpret the data and what action is warranted. However, we all agree that the question of CV risk associated with febuxostat is complicated and is not cleanly resolved at this point despite completion of the CARES study, a well-controlled trial in a high-risk CV population. In addition, we agree that the Agency should decide at what point we should cease further investigations into CV safety signals considering the inconsistent results obtained to date. Our division director, Dr. Norman Stockbridge, concludes that accumulated cardiac safety data from febuxostat clinical studies, including CARES, do not provide conclusive evidence that febuxostat use is associated with an increased risk for CV death. Furthermore, Dr. Stockbridge believes it is premature to recommend changes to product labeling regarding CV death risk without more conclusive data.

Takeda provided requested information but no data pertaining to their ongoing European postmarketing CV safety study, “Febuxostat versus Allopurinol Streamlined Trial (FAST): A Prospective, Randomized, Open-Label, Blinded Endpoint (PROBE) Clinical Trial Evaluating the Long Term Cardiovascular Safety of Febuxostat in Comparison with Allopurinol in Patients with Chronic Symptomatic Hyperuricemia.” We believe that data obtained from the FAST trial may contribute to a more robust safety database. However, because of differences in CV risk among enrolled subjects and febuxostat dosing, FAST may not yield helpful supportive evidence for whether there is or is not increased risk for CV death associated with febuxostat.

Background

Uloric (febuxostat) is a xanthine oxidase inhibitor (XOI) indicated for chronic management of hyperuricemia in patients with gout. The Agency approved Uloric (febuxostat) 40 mg and 80 mg tablets in 2009 after two prior NDA submission cycles in 2004 and 2006. Takeda conducted two pivotal Phase 3 trials in support of NDA approval. The first study, “Febuxostat Versus Allopurinol Controlled Trial” (Study C02-010, FACT), was a randomized controlled trial in gout patients without significant renal impairment randomized to either febuxostat (80 mg or 120 mg) or active comparator (allopurinol) for 52 weeks. Out of 760 subjects, 8 deaths were reported with either 80 mg or 120 mg febuxostat doses compared to no deaths in the allopurinol arm. Two of these deaths were attributed to myocardial infarction. The second study, “The Allopurinol- and Placebo-Controlled Efficacy Study of Febuxostat” (C02-009, APEX) was a randomized controlled trial in gout patients with normal or impaired renal function randomized to receive febuxostat (80 mg, 120 mg, or 240 mg), active comparator (allopurinol), or placebo for up to 28 weeks. Out of 1062 subjects, there were no deaths reported but there were numerically more CV serious adverse events (SAEs) in subjects who received febuxostat compared to allopurinol and placebo. Of note, in both studies, the proportion of subjects enrolled with pre-existing CV disease ranged from 7 to 18%.

Considering all Phase 2 and 3 study data included in the original NDA submission, DPARP concluded there was a significant imbalance in CV SAEs reported with febuxostat compared to allopurinol for thrombotic events including ischemic coronary artery disease, cerebrovascular accident, and transient ischemic attack in addition to congestive heart failure. Study limitations of the two pivotal Phase 3 trials, namely small numbers of CV adverse events (AEs), made it difficult for DPARP to reach firm conclusions about the significance of cardiovascular risk associated with febuxostat. DPARP did not approve either the 2004 original NDA or 2006 NDA resubmission because of persistent concern for CV safety imbalance including all-cause death, cardiovascular mortality, and

100 of 169 2

Reference ID: 4358691 thromboembolic SAEs associated with febuxostat. Of note, preclinical studies did not demonstrate CV safety or CV toxicity concerns.

In 2008, Takeda resubmitted NDA 21856 for febuxostat tablets containing a new six-month Phase 3 randomized controlled trial (F-GT06-153) to evaluate febuxostat 40 and 80 mg in comparison to allopurinol in subjects with gout. Takeda designed the study to determine whether a cardiovascular safety signal seen in prior studies would be observed again in a larger study and in one with a predefined and adjudicated CV endpoint, Anti-Platelet Trialists’ Collaboration (APTC) events. Out of 2269 enrolled subjects, about 1300 had a prior history of cardiac disease. Three hundred nineteen (319) potential CV events were adjudicated in a blinded fashion with three (3) adjudicated CV events each reported in febuxostat 80 mg (3/756, 0.3%) and allopurinol (3/756, 0.4%) arms. There were numerically more all-cause deaths in the allopurinol arm (3/756, 0.4%) compared to febuxostat [40 mg: 1/757 (0.1%) and 80 mg: 1/756 (0.1%)]. The two reported CV deaths occurred in the allopurinol arm (2/756, 0.3%). DPARP concluded that the overall incidence of CV APTC events was not increased in febuxostat-treated patients compared to those treated with allopurinol. Of adjudicated non-APTC events including unstable angina, coronary revascularization, transient ischemic attack (TIA), cerebral vascularization, congestive heart failure, and arrhythmia, there were numerically more events in febuxostat treatment arms, 40 mg (10/757, 1.3%) and 80 mg (9/756, 1.2%) compared to the allopurinol arm (7/756, 0.9%) but without obvious dose-dependent increases in events. Of note, congestive heart failure was reported in 2/757 (0.3%) in the febuxostat 40 mg arm compared to zero (0) and 1/756 (0.1%) in febuxostat 80 mg and allopurinol arms, respectively.

Prior to approval of Uloric (febuxostat) tablets in 2009, DPARP presented safety and efficacy data for febuxostat at an Advisory Committee Meeting in 2008 resulting in a 12-0 vote (1 abstention) in favor of NDA approval. At time of approval, drug labeling included a CV Warning and Precaution regarding a higher rate of thromboembolic events observed in patients treated with febuxostat. In addition, DPARP issued a post-marketing requirement (PMR) to conduct a CVOT. In Jan 2018, Takeda submitted a supplemental NDA (NDA 21856/S- 013), the subject of this review, containing their final study report and data analyses from their CVOT, “A Multicenter, Randomized, Active Control, Phase 3B Study to Evaluate the Cardiovascular Safety of Febuxostat and Allopurinol in Subjects with Gout and Cardiovascular Comorbidities” (CARES). DPARP requested that DCaRP review and comment on the CVOT study design, endpoints, adjudication process, and results. DPARP previously consulted DCaRP in 2006 and again in 2008 regarding appropriateness of CV event assessment methods and evaluation of reported CV AEs in prior NDA submissions for febuxostat tablets.

Review of Study TMX-67-301 (CARES)

Study Title: A Multicenter, Randomized, Active-Control, Phase 3B Study to Evaluate the Cardiovascular Safety of Febuxostat and Allopurinol in Subjects with Gout and Cardiovascular Comorbidities

Objective: Compare the risk of predefined Major Adverse Cardiovascular Events (MACE) during treatment with febuxostat and allopurinol in subjects with gout and CV comorbidities

Primary Endpoint: Time from randomization to first occurrence of any event in the predefined MACE composite, including CV death, nonfatal MI, nonfatal stroke, and unstable angina with urgent coronary revascularization

Other Endpoints Pertinent to this Review:  Time from randomization to first occurrence of any Antiplatelet Trialists’ Collaborative (APTC) event including CV death, nonfatal MI, and nonfatal stroke  Time from randomization to occurrence of each individual event in the predefined MACE composite  Time from randomization to first occurrence of any event in the predefined MACE composite plus any of the following events: – urgent cerebral revascularization (nonelective) – hospitalized congestive heart failure (CHF) – arrhythmias not associated with ischemia

101 of 169 3

Reference ID: 4358691 – venous and peripheral arterial thromboembolic events; e.g., deep vein thrombosis (DVT) – pulmonary embolism (PE) – transient ischemic attack (TIA)

Key Eligibility Criteria:  Male ≥50 years of age or female ≥55 years of age, postmenopausal  History or presence of gout  History of major CV or cerebrovascular disease including at least 1 of the following: – MI – hospitalized unstable angina – cardiac or cerebrovascular revascularization procedure – stroke – hospitalized transient ischemic attack (TIA) – peripheral vascular disease – history of diabetes mellitus with evidence of microvascular or macrovascular disease  No history of MI or stoke within 60 days prior to Screening Visit

Study Design: Study TMX-67-301 was a double-blind and active-controlled study. Subjects were randomized 1:1 to receive either febuxostat or allopurinol daily. Subjects randomized to febuxostat initially received 40 mg daily with uptitration to 80 mg if serum uric acid (sUA) remained ≥ 6.0 mg/dL by the Week 4 Visit. Subjects randomized to allopurinol initially received either 200 mg or 300 mg daily depending on baseline renal function with predefined uptitration to a max dose of either 400 mg or 600 mg depending on sUA. Subjects were stratified according to baseline renal function, i.e., normal/mild renal impairment versus moderate renal impairment. Study duration varied because the study was event-driven for a maximum of 624 MACE. A Data Monitoring Committee (DMC) periodically reviewed unblinded study results to monitor study progress and subject safety.

Adjudication Process: Takeda utilized a clinical endpoint committee (CEC) to adjudicate all potential CV SAEs and deaths. The CEC consisted of three members with expertise in cardiovascular or cerebrovascular disease. After a subject reported a SAE, investigators completed a SAE form within 1 business day of initial onset or notification. Takeda designated an adjudication team, “PPD Adjudication Team,” to send a listing of all new SAEs, approximately every month or sooner, to the CEC Chair to select appropriate events for committee review. After the CEC chair selected potential events for adjudication, the PPD Adjudication team worked with study sites to compile forms and supporting documentation for non-fatal CV SAEs or deaths for distribution to CEC members using an electronic software database program. CEC members received electronic alerts notifying them that a package was ready for their individual review and adjudication. Discordant adjudications were flagged by the electronic database thereby requiring the committee to review these cases together and make an adjudication determination by simple majority vote. CEC members were provided pre-specified definitions for MACE and non-MACE CV events.

Statistical Plan Overview: CARES was an event-driven trial designed for a maximum of 624 events to assess noninferiority of febuxostat compared to allopurinol assuming a true hazard ratio (HR) of 1.0 and 90% power. Noninferiority was assessed only for the primary safety endpoint, MACE composite. Noninferiority of febuxostat to allopurinol was declared (at an interim or final analysis) if the upper 1-sided confidence interval (CI) for the HR, while preserving an overall 1-sided false-rejection rate of 2.5%, was less than 1.3. To identify risk factors for MACE and APTC events, the sponsor performed additional analyses using multivariate Cox Proportional Hazards modeling. Interim analyses were conducted by an independent statistician not involved in the conduct of the study when approximately 25%, 50%, and 75% of events occurred followed by a final analysis. At each analysis, if the upper confidence limit of the HR was ≤1.3, the study was to be stopped and the

102 of 169 4

Reference ID: 4358691 noninferiority of febuxostat relative to allopurinol for CV risk was to be declared. If non-inferiority was declared at an interim analysis, subsequent analyses of the primary endpoint were considered sensitivity analyses.

Reviewer Comments:  The study design, primary and secondary endpoints, and adjudication process are acceptable to assess CV safety in a gout population with moderate to high CV risk. In Takeda’s pivotal Phase 3 trials, studies C02- 009 and C02-010, < 20% of enrolled subjects had baseline cardiovascular disease.  The criterion for noninferiority of febuxostat compared to allopurinol was met at the 75% interim analysis for the primary endpoint. However, the DMC recommended that Takeda continue the study until the maximum 624 events occurred, to which Takeda agreed.

Key Study Safety Results: According to summary demographic tables provided by Takeda, baseline demographics including CV history were similar between febuxostat and allopurinol treatment arms. For the primary endpoint, time from randomization to first appearance of any event in the predefined MACE composite, febuxostat was noninferior to allopurinol with the upper bound of the CI for the HR 1.23. However, there was a nominally statistically significant imbalance in CV death in the febuxostat treatment arm [HR 1.34, CI (1.03, 1.73)], also reflected in an imbalance in sudden cardiac death in the febuxostat treatment arm. Takeda’s subgroup analyses were unable to identify an explanation. All-cause and CV death results are described in more detail below.

Deaths: Rates for all-cause and adjudicated CV death were higher with febuxostat [n=243/3098 (7.8%) and n=134/3098 (4.3%), respectively] compared to allopurinol [n=199/3092 (6.4%) and n=100/3092 (3.2%), respectively]. The imbalance of adjudicated CV death in the febuxostat arm was also observed in sensitivity analyses. Of all adjudicated causes of CV death, more sudden cardiac death events occurred in the febuxostat treatment arm (n=83/3098, 2.7%) compared to allopurinol (n=56/3092, 1.8%). Takeda conducted subgroup analyses for CV death based on demographic, baseline characteristics, and post-baseline characteristics. The only covariates that appeared to contribute to increased risk for all-cause death or CV death were no use of low- dose aspirin and use of non-steroidal anti-inflammatories (NSAIDs) or colchicine [relative risks (RRs) between 1.7 to 2.0]. Otherwise, Takeda could not identify an obvious subgroup population or baseline characteristic associated with increased risk for CV death with febuxostat. Adjudicated causes of death are shown in the next section.

Reviewer Comments: In the absence of significant imbalances in other CV safety signals associated with febuxostat, the clinical significance of CV death is unclear. Moreover, despite various subgroup analyses, Takeda could not identify a plausible risk factor to explain an increased risk for cardiac death with febuxostat treatment. We do not believe that concomitant medication use or not of aspirin, NSAIDs, or colchicine provides a plausible explanation for CV death imbalance with febuxostat treatment.

Cardiovascular Safety Data Summary

To explore potential mechanisms/causes of the observed increase in CV death in the febuxostat arm versus allopurinol in CARES, we examined nonclinical and clinical evidence for a febuxostat-related effect on the following events: heart failure, thromboembolic/ischemic heart disease, and arrhythmia.

Nonclinical Studies

Overall, animal models studying febuxostat did not indicate any direct effect on myocardial function or hemodynamics which would suggest a harmful mechanism. In addition, animal models studying febuxostat did not indicate propensity for febuxostat to affect coagulation or platelet aggregation. In vitro and in vivo studies did not show significant effects of febuxostat on sodium, potassium, or calcium channels, even at supratherapeutic concentrations. In vitro studies provided supportive evidence that febuxostat would not have

103 of 169 5

Reference ID: 4358691 the adverse effect of Torsades de pointes. In vivo studies in dogs with febuxostat up to 100 mg/kg (15 times higher than therapeutic dose) did not show effects on cardiovascular parameters including ECG changes.

The sponsor conducted the following nonclinical studies pertinent to our review:

 The initial NDA submission included studies of the effects of TMX-67 on: o Action potential parameters in dog isolated cardiac purkinje fibers o Human cardiac sodium currents expressed in mammalian cells o Cloned hERG channels expressed in mammalian cells o Cloned hERG channels expressed in Chinese hamster ovary cells o Cardiovascular findings in beagle dogs o Human platelets and blood coagulation o Action potentials in isolated cardiac purkinje fibers o Native cardiac L-type calcium current of guinea-pig cardiomyocytes o The antihypertensive activity of nifedipine in spontaneous hypertensive rats

 3rd NDA submission – additional nonclinical studies submitted o Prevention and treatment effects of xanthine oxidase inhibition on systemic overload induced ventricular hypertrophy (LVH) and congestive heart failure (CHF) in mice . LVH and CHF were induced in mice by aortic constriction and subsequently treated with febuxostat. There were 2 treatment groups: treatment initiated shortly post-op/high dose (5 mg/kg/day oral x 8 days) and late treatment initiation at 7 days post-op/low- dose (0.05 mg/kg/day oral x 21 days) to assess the ability of febuxostat to prevent CHF. ECG, plasma uric acid, histopathology of heart, collagen content, and “cardiac protein levels” were evaluated. In the early treatment/high-dose group, the severity of LVH and induction of “several heart proteins” were reduced. In the delayed treatment/low-dose group established LVH and dysfunction did not improve. The applicant concluded that inhibition of xanthine oxidase by febuxostat prevented the development of cardiac hypertrophy and CHF in mice but was not effective in treatment of established cardiac hypertrophy and dysfunction. o Effect of acute xanthine oxidase inhibition with febuxostat or allopurinol on myocardial energetics during basal and high cardiac work-states . The sponsor investigated the effect of febuxostat and allopurinol on catecholamine- induced myocardial workload by measuring myocardial blood flow, high energy phosphate level and oxygen consumption in open chest instrumented dogs. The dose of febuxostat, 4 mg/kg, given as a slow IV infusion for 30 minutes showed a lowering of free ADP suggesting greater utilization of myocardial energy due to inhibition of oxidative states.

Adverse Event Analyses

Heart Failure

Treatment Emergent Adverse Events (TEAE) in CARES

An analysis by the Cardiac Failure SMQ (narrow) did not reveal any important differences in incidence of events between febuxostat and allopurinol (4.9 per 100 pt-yrs vs 4.7 per 100 pt-yrs). The predominant preferred term (80%) in this SMQ was “cardiac failure congestive” and the incidence was not different between groups. An analysis by the broad Cardiac Failure SMQ was consistent in that no relevant differences were noted between groups. The broad SMQ mostly added terms related to edema (“Oedema” and “Oedema peripheral”) for which there was no difference between groups.

104 of 169 6

Reference ID: 4358691 Serious TEAEs in CARES Similarly, there was little difference between groups in an analysis of serious TEAEs in the Cardiac Failure SMQ (narrow) (febuxostat 3.6 per 100 pt-yrs vs allopurinol 3.2 per 100 pt-yrs). Very few serious edema events were reported (4 patients in each group).

Thromboembolic/Ischemic Heart Disease Events

TEAEs in CARES

An analysis by the Embolic/thrombotic event SMQ (narrow) showed a very small difference in the incidence of events between febuxostat and allopurinol (6.1 per 100 pt-yrs vs 5.7 per 100 pt-yrs). A review of terms within this SMQ revealed no consistent or important differences in preferred terms representing thromboembolic events (myocardial infarction, deep vein thrombosis). There was little difference between groups in an analysis of events reported in the (narrow) Ischemic Heart Disease SMQ (febuxostat 7.8 per 100 pt-yrs vs allopurinol 8.0 per 100 pt-yrs), consistent with the broad SMQ analysis.

Serious TEAEs in CARES

There was little difference between groups for the Embolic/thrombotic SMQ (narrow) in an analysis of serious TEAEs (febuxostat 4.8 per 100 pt-yrs vs allopurinol 4.5 per 100 pt-yrs) or in an analysis of serious TEAEs in the Ischemic Heart Disease SMQ (febuxostat 5.6 per 100 pt-yrs vs allopurinol 5.7 per 100 pt-yrs).

Arrhythmias

TEAEs in CARES

There was little difference between groups in an analysis of events in the Cardiac Arrhythmia SMQ (febuxostat 2.5 per 100 pt-yrs vs allopurinol 2.4 per 100 pt-yrs). Additional events analyzed in the broad SMQ analysis were not informative. A review of commonly reported individual events revealed similar numbers of ventricular events (febuxostat 55 vs allopurinol 57) and atrial fibrillation/flutter (170 vs 165) between groups. A review of other arrhythmic events (blocks, bradycardias) did not reveal any signals.

Serious TEAEs in CARES There is little difference between groups in an analysis of serious TEAEs in the Cardiac Arrhythmia SMQ (narrow) (febuxostat 2.5 per 100 pt-yrs vs allopurinol 2.4 per 100 pt-yrs).

Death Analyses

Of the 442 deaths that occurred in the CARES study, 234 were adjudicated as CV death and 208 were adjudicated as non-CV death. Adjudicated causes of CV death in CARES are listed in Table 1 below.

105 of 169 7

Reference ID: 4358691 Table 1: Adjudicated Cause of CV Death (CARES)

Source TMX-67_301 Study Report Table 11.m

Reviewer Comments:  Takeda provided a summary of baseline CV history of subjects in the CARES study. Overall, baseline CV history including congestive heart failure were similar between febuxostat and allopurinol treatment arms.  Review of Takeda’s MedDRA data grouped by cardiac SMQs as opposed to only preferred terms did not elucidate an obvious explanation for the CV death imbalance with febuxostat.

106 of 169 8

Reference ID: 4358691 Signature Page 1 of 1 ------This is a representation of an electronic record that was signed electronically. Following this are manifestations of any and all electronic signatures for this electronic record. ------/s/ ------

SHETARRA E WALKER 12/04/2018

MARY R SOUTHWORTH 12/06/2018

MARTIN ROSE 12/10/2018

NORMAN L STOCKBRIDGE 12/10/2018

107 of 169

Reference ID: 4358691 Department of Health and Human Services Public Health Service Food and Drug Administration Center for Drug Evaluation and Research Office of Surveillance and Epidemiology (OSE) Office of Pharmacovigilance and Epidemiology (OPE)

Epidemiology literature and drug utilization review

Date: 11/21/2018

Reviewer(s): Marie Bradley, PhD, MSc.PH, MPharm Division of Epidemiology II Kusum Mistry, Pharm.D, Drug Utilization Analyst Division of Epidemiology II Team Leader(s): Efe Eworuke, PhD Division of Epidemiology II LCDR Justin Mathew, Pharm.D., USPHS Team Leader for Drug Utilization Division of Epidemiology II Division Director(s): Lockwood Taylor, PhD Division of Epidemiology II LCDR Grace Chai, PharmD, USPHS, Deputy Director for Drug Utilization Division of Epidemiology II

Subject Cardiovascular risk and mortality in patients with gout using febuxostat compared to allopurinol. Drug Name(s): Uloric (febuxostat) Application Type/Number: NDA 021856 Applicant/sponsor: Takeda USA OSE RCM #: 2018-1542 TSI #: 1849

108 of 169 TABLE OF CONTENTS

EXECUTIVE SUMMARY ...... 2 1 INTRODUCTION ...... 3 1.1 Background ...... 3 1.2 Regulatory History ...... 4 1.3 Product Labeling ...... 5 2 REVIEW METHODS AND MATERIALS ...... 5 2.1 Literature search strategy ...... 5 2.2 Drug Utilization Methods...... 6 2.2.1 Data Sources Used ...... 6 2.2.1.1 U.S. Sales Distribution Data ...... 6 2.2.1.2 U.S Outpatient Retail Pharmacy Data ...... 7 2.2.1.3 U.S Office-Based Physician Survey Data ...... 8 3 REVIEW RESULTS ...... 8 3.1 Literature Search Results ...... 8 3.2 Drug Utilization Results ...... 15 3.2.1 Sales Distribution Data ...... 15 3.2.1.1 Settings of Care ...... 15 3.2.2 U.S. Outpatient Retail Pharmacy Data ...... 15 3.2.2.1 Prescription-Level Data ...... 15 3.2.2.2 Prescription-Level Data by Drug Strength ...... 16 3.2.2.3 Patient-Level Data by Sex and Age Groups ...... 16 3.2.2.4 Prescriber Specialty Data ...... 17 3.2.3 U.S. Office-Based Physician Survey Data ...... 17 3.2.3.1 Survey Data by Diagnoses and Total Daily Dose ...... 17 3.2.3.2 Survey Data by Diagnoses and Directions of Use (Signa) ...... 17 4 DISCUSSION ...... 17 5 CONCLUSIONS ...... 19 6 RECOMMENDATIONS...... 20 7 REFERENCES ...... 20 8 APPENDICES ...... 21 8.1 Drug Utilization Database Descriptions and Limitations ...... 22 8.2 Drug Utilization Tables ...... 24

109 of 169 EXECUTIVE SUMMARY Febuxostat (Uloric®) is a selective inhibitor of xanthine oxidase that lowers serum uric acid levels by inhibiting the conversion of xanthine to uric acid. It was approved by the FDA in February 2009, for the management of chronic hyperuricemia in patients with gout, following three review cycles that included an arthritis advisory committee (AC) meeting, which focused on the drug’s cardiovascular (CV) safety. As a condition of approval, the FDA required a post-market randomized controlled trial (RCT) to determine whether febuxostat use was associated with an increase in the risk of serious CV outcomes compared to allopurinol. The preliminary results from this study, known as The Cardiovascular Safety of Febuxostat and Allopurinol in Patients with Gout and Cardiovascular Morbidity (CARES) study, released in 2017, showed an increased risk of cardiovascular-related death and all-cause death in febuxostat users. As a result, FDA issued a drug safety communication for febuxostat in November 2017 (https://www.fda.gov/Drugs/DrugSafety/ucm584702.htm), and in June, 2018, Public Citizen submitted a Citizen’s Petition requesting the immediate removal of febuxostat from the US market based on the results from the CARES study. The Division of Epidemiology (DEPI) received a consult request from the Division of Pulmonary, Allergy, and Rheumatology Products (DPARP) to conduct a literature review to identify any observational studies examining CV risk and mortality in patients with gout using febuxostat compared to allopurinol, in real world settings and to provide drug utilization data for urate lowering therapies (ULT).

To characterize drug utilization patterns for ULTs in the U.S., analyses using outpatient retail pharmacy data and physician office-based physician survey data were conducted. Outpatient retail pharmacy data showed that allopurinol was the most widely used ULT with 14.9 million prescriptions dispensed compared to 1.2 million for febuxostat, in 2017. Nationally estimated prescription data suggested the use of allopurinol and febuxostat were increasing between 2013-2017. The most commonly dispensed strengths were allopurinol 300 mg tablets and febuxostat 40 mg tablets. Utilization by sex and age found that males and patients aged 65 years or older accounted for the largest proportion of allopurinol and febuxostat use.

Four relevant abstracts were identified from searches in PubMed and Embase. Two were ultimately excluded; one combined febuxostat and allopurinol in the same exposure group and the other was an abstract from conference proceedings. The remaining two were selected for full text examination and subsequent inclusion in the review.

One of the included studies, by Foody et al, had several limitations that made it challenging to meaningfully interpret the study results for this review. Therefore, the focus was on the study by Zhang et al. Unlike the CARES trial, the study by Zhang et al found that among U.S. Medicare beneficiaries with gout, there was no difference in risk of all-cause mortality between febuxostat and allopurinol users in both the main analysis and in subgroup analyses restricted to those with high baseline CVD risk. These results were observed in the as treated as well as intention-to-treat analyses. Despite this, when the main analysis was stratified by follow-up time, a trend toward a higher risk of all-cause mortality was seen in febuxostat users with more than 3 years of follow-up, although statistical significance was not attained. The study also reported that initiation of febuxostat compared with allopurinol was not associated with an increased risk of CV events including myocardial infarction (MI), stroke, and coronary revascularization. A modestly decreased risk for heart failure exacerbation associated with febuxostat versus allopurinol use was reported.

The study by Zhang et al was well conducted using an active comparator to reduce unmeasured confounding, propensity score matching to balance comparator cohorts on known potential confounders, validated algorithms for outcome ascertainment and both as treated and ITT type analyses to reflect risk among those currently on treatment and risk among those who recently discontinued. However, a

110 of 169 prevalent new user design was applied in the main analysis and sub-group analyses. This design can lead to issues with selection bias, immortal time bias and potentially adjusting for intermediate risk factors on the causal pathway. Despite this, the authors conducted a sensitivity analysis in which users were naive to both drugs in the 365 days prior to initiation, although not on the entire study cohort, and the results were consistent with the main analyses.

DEPI has no recommendations at this time. The need for further CV safety data from long-term users of febuxostat and allopurinol should be addressed in the FAST trial, currently underway in the UK and Denmark.

1 INTRODUCTION Febuxostat (Uloric®) is a selective inhibitor of xanthine oxidase that lowers serum uric acid levels by inhibiting the conversion of xanthine to uric acid. It was initially approved by the FDA in February 2009 at doses of 40 mg and 80 mg once daily following three review cycles that included an arthritis advisory committee (ACC) meeting which focused on the drug’s cardiovascular (CV) safety. As a condition of approval, the FDA required a post-market randomized controlled trial (RCT) to determine whether the use of febuxostat is associated with a moderate increase in the risk of serious adverse CV outcomes as compared to allopurinol, known as the CARES study, which started in 2010.The preliminary results from the CARES study, issued in 2017 showed an increased risk of cardiovascular-related death and all-cause death with febuxostat compared with allopurinol. As a result, FDA issued a drug safety communication for febuxostat in November 2017 (https://www.fda.gov/Drugs/DrugSafety/ucm584702.htm) and in June, 2018, Public Citizen submitted a Citizen’s Petition requesting the immediate removal of febuxostat from the US market based on the results from the CARES study.

The Division of Epidemiology (DEPI) received a consult request from the Division of Pulmonary, Allergy, and Rheumatology Products (DPARP) to conduct a literature review to identify any observational studies examining CV risk and mortality in patients with gout using febuxostat compared to allopurinol, in real world settings. DPARP also requested that DEPI provide drug utilization data for ULT.

1.1 BACKGROUND Gout is a metabolic disease characterized by uric acid crystal deposition in the joints, the kidneys, and other tissues that may lead to locally destructive tophi as well as uric acid nephrolithiasis and interstitial kidney disease.1 Gout typically manifests as an inflammatory arthropathy including chronic monoarthritis or polyarthritis punctuated by episodic painful flares commonly affecting the metatarsophalangeal joint of the first toe, as well as the tarsal joints, ankles, and knees2-5.

Febuxostat is a selective inhibitor of xanthine oxidase (XOI) that lowers serum uric acid levels by inhibiting the conversion of xanthine to uric acid. It was approved, in 2009, at doses of 40 mg and 80 mg once daily for the management of chronic hyperuricemia in patients with gout. Urate lowering therapy (ULT) is central to long-term management of gout. Some evidence suggests that high levels of uric acid represent an independent risk factor for cardiovascular disease (CVD) and that the use XOI may reduce the risk of major adverse cardiovascular events6-8.

A number of pre-approval RCTs (FACT, APEX) submitted to FDA as part of the first NDA submission (NDA 21-856) in December 2004 in support of febuxostat’s safety and efficacy, suggested febuxostat use led to an increased risk of CV events and death. At the end of the first review cycle, FDA determined that subjects who had received febuxostat had a higher rate of overall mortality, mortality from CV causes, and a higher rate of CV thromboembolic events. An approvable letter was issued on October 14, 2005,

111 of 169 and additional safety data was required. In February 2006, a second review cycle was triggered based on reanalysis of the prior clinical trial data augmented by new data from the two then-ongoing long-term extension RCTS (FOCUS and EXCEL). FDA were still concerned about increased risk of all-cause mortality, CV mortality, and serious CV events with febuxostat and asked the sponsor to conduct “A Randomized, Multicenter Study Comparing the Efficacy and Safety of febuxostat to Allopurinol in Reducing the Incidence of Gout Flares in Subjects with Gout.” The results of this trial known as the CONFIRMS trial were published in 2009 and showed no safety signal for adverse CV events, but the upper bounds of the 95% CIs indicated the results could not exclude an increased risk. A third NDA submission was initiated in July 2008 after completion of the CONFIRMS trial for approval of febuxostat 40-mg and 80-mg tablets. Uncertainty at the agency remained about the CV safety of febuxostat, and an arthritis Advisory Committee (AC) was convened on November 24, 2008, which voted 12 to 0, with 1 abstention, to recommend FDA approval of the drug, with a requirement for post market studies to further assess the drug’s safety. Febuxostat was subsequently approved on February 13th, 2009. The post market RCT known as The Cardiovascular Safety of Febuxostat and Allopurinol in Patients with Gout and Cardiovascular Morbidity (CARES) study, was a multicenter, randomized, double-blind, CV outcome safety study conducted in the U.S., Canada and Mexico comparing febuxostat to allopurinol among those with existing CV disease. In October 2017, the sponsor (now Takeda) informed FDA of the results of the preliminary results from CARES which showed a significant increase in CV and all-cause mortality in the febuxostat arm compared to the allopurinol arm but no increase in CV events. On November 15, 2017, FDA issued a Drug Safety Communication about the risk of CV death and all-cause death with febuxostat compared to allopurinol https://www.fda.gov/Drugs/DrugSafety/ucm584702.htm found in the CARES study.

The European Medicines Agency (EMA) also instructed the sponsor to conduct a post marketing study to evaluate CV safety of febuxostat as part of a pharmacovigilance plan for the drug. The Febuxostat versus Allopurinol Streamlined Trial (FAST), a prospective, randomized, open-label, blinded endpoint (PROBE) clinical trial evaluating the long-term CV safety of febuxostat in comparison with allopurinol in patients with chronic symptomatic hyperuricemia, when taken for an average of 3 years in patients aged 60 years or older, is currently ongoing in the UK and Denmark 9.

On June 21, 2018 a Citizen’s Petition was submitted to FDA by Public Citizen. The petition requests that FDA remove all drug products containing febuxostat from the market because (1) febuxostat use increases the risk of death compared with alternative therapies and (2) there exist other effective medications that have been approved by the FDA for treatment of gout that have a lower risk of death. The petitioner asserted that febuxostat has unique serious risks but no unique clinical benefit.

However, given the limited number of ULTs available, the growing gout epidemic, and the unmet medical need in the gout population (7% of all races suffer allopurinol hypersensitivity syndrome) febuxostat may still have a role in the management of gout.

Regulatory History Pertinent regulatory milestones are presented below:

December 2004: First NDA 21856 submitted.

October 14 2005: Approvable letter issued citing “concerns regarding the potential for (febuxostat) to cause clinically significant cardiovascular/thrombotic adverse events in excess to that seen with allopurinol or placebo, even when exposure -over-time is factored into the analysis” along with a request for additional information regarding the cardiovascular safety of febuxostat derived from “comparative controlled safety data, or possibly through re-analyses of the current database.”

112 of 169

February 22, 2006: The second review cycle was initiated.

August 2, 2006: A second approvable letter was issued which stated that the Applicant had to “provide further data to clarify the cardiovascular risks of the proposed doses and/or provide data on the safety and efficacy of lower doses of febuxostat in order to assure us that a dose level(s) with favorable risk- benefit characteristics has been defined.” It was also suggested that the Applicant conduct a new study designed to determine whether the cardiovascular safety signal was real.

June 2008: Third and final review cycle triggered when the applicant submitted efficacy and safety data from a new phase 3 RCT, the CONFIRMS trial (F-GT06-53), as well as an additional 12 months of exposure data from the two long term extension studies, TMX-01-005 and C02-021

November 24, 2008: Arthritis AC meeting held, and the panel voted 12 to 0 with 1 abstention, to recommend approval of febuxostat with a post-marketing safety study to further assess the CV risk associated with the drug. February 13, 2009: Febuxostat approval letter issued.

November 2016: Takeda informed DPARP of an increase in all-cause mortality (due to CV mortality) in the febuxostat arm of CARES trial.

December 2016: DPARP sent Takeda a correspondence stating they would not review the interim data or interfere with the conduct of the study.

March 2017: Takeda informed DPARP that they were allowing the study to continue to reach the targeted number of MACE.

October 2017: Takeda informed FDA of the results of the primary analysis from the CARES study which showed a statistically significant increase in CV death and all-cause death in febuxostat users compared to allopurinol users.

November 15, 2017: FDA issued a Drug Safety Communication about the increased risk of CV and all- cause mortality seen in febuxostat users in the CARES study.

January 19, 2018: The final CARES study report and supplement were submitted to FDA.

June, 2018: Public Citizen submitted a Citizen’s Petition requesting the immediate removal of febuxostat from the US market

1.2 PRODUCT LABELING

------5 WARNINGS AND PRECAUTIONS------Cardiovascular Events: A higher rate of cardiovascular thromboembolic events was observed in patients treated with ULORIC than allopurinol in clinical trials. Monitor for signs and symptoms of MI and stroke.

2 REVIEW METHODS AND MATERIALS

2.1 LITERATURE SEARCH STRATEGY

113 of 169 Step 1: The following combination of search terms were applied in both PubMed (National Library of Medicine, Bethesda, MD) and Embase (Reed Elsevier PLC, Amsterdam, The Netherlands) to identify observational studies related to febuxostat use and CV events and mortality. The search ran from database inception to October 15th, 2018. The restrictions of English language and human only studies were applied.

1. “Febuxostat” AND “Cardiovascular disease” PubMed 117 results, Embase 681 results 2. “Febuxostat” AND “CVD” PubMed 7 results, Embase 24 3. “Febuxostat” AND “allopurinol” AND “cardiovascular disease” PubMed 80 results, Embase 190 4. “Febuxostat” AND “Mortality” PubMed 28 results, Embase 106 5. “Febuxostat” AND “allopurinol” AND “Mortality” PubMed 80 results, Embase 88

Step 2: Article titles and abstracts were screened (by inclusion criteria below) to assess relevancy for full text examination by DEPI’s reviewer. Step 3: For the publications identified in Step 2, reference sections were reviewed for additional studies.

Inclusion Criteria The following inclusion criteria were applied to select appropriate studies: • Population: adults aged 18 years and older • Exposure: febuxostat • Comparator: allopurinol • Outcome: CV events and mortality or all-cause mortality • Study type: Observational studies including: cohort, case control, cross sectional

2.2 DRUG UTILIZATION METHODS

2.2.1 Data Sources Used Proprietary drug utilization databases available to FDA were used to conduct this analysis. Detailed descriptions and limitations of the databases are included in Appendix 8.1.

2.2.1.1 U.S. Sales Distribution Data The IQVIA™ National Sales Perspectives (NSP) database was used to determine the settings of care where febuxostat, allopurinol, and other ULTs (probenecid, rasburicase and pegloticase) were sold by the manufacturer to various U.S. channels of distribution in 2017. Of note, IV rasburicase and IV pegloticase are primarily administered in the hospital and clinic settings.a

a IQVIA™ National Sales Perspective. 2017. Extracted October 2018. File: NSP 2018-1544 febuxostat and other molecules by Sup Ch 10-30-2018.

114 of 169 Therefore, our outpatient prescription analysis included only utilization patterns for febuxostat, allopurinol, and probenecid.

2.2.1.2 U.S Outpatient Retail Pharmacy Data The IQVIA™ National Prescription Audit (NPA) database was used to obtain the nationally estimated number of prescriptions dispensed for febuxostat, allopurinol, or probenecid from U.S. outpatient retail pharmacies (for any indication) from 2013 through 2017. In addition, NPA was used to obtain the nationally estimated number of prescriptions dispensed for febuxostat or allopurinol, stratified by drug strength for the same time-period and the top prescriber specialties in 2017. The IQVIA™ Total Patient Tracker (TPT) database was used to obtain the nationally estimated number of patients who received a dispensed prescription for febuxostat or allopurinol (prescribed for any indication) from U.S. outpatient retail pharmacies, stratified by patient sex and age groups (≤20 years, 21- 44, 45-64 years, and 65 years or older) from 2013 through 2017. Although all formulations (oral and injectable) of allopurinol were included in our utilization analysis, the number of prescriptions dispensed for the injectable formulation accounted for a negligible amount in the outpatient retail setting.b,c

b IQVIA™ National Prescription Audit. 2013-2017. Extracted October 2018. File: NPA 2018-1544 febuxostat and allopurinol PFS 10-29-2018 c IQVIA™ Total Patient Tracker. 2013-2017. Extracted October 2018. File: TPT 2018-1544 febuxostat and allopurinol PFS Retail TRx 10-29-2018

115 of 169 2.2.1.3 U.S Office-Based Physician Survey Data The Syneos Health Research and Insights Treatment Answers™ was used to examine the top diagnoses, total daily dose, and directions for use (signa) associated with the use of febuxostat or allopurinol from U.S. office-based physician surveys in 2017. Diagnoses data by the number of drug use mentionsd were captured based on International Classification of Diseases, Tenth Revision, Clinical Modification (ICD- 10-CM) codes with 95% confidence intervals.

3 REVIEW RESULTS

3.1 LITERATURE SEARCH RESULTS

Four abstracts were screened as potentially eligible for full text examination. Two were ultimately excluded; one combined febuxostat and allopurinol in the same exposure group10 and the other was an abstract from conference proceedings. The remaining two study abstracts 11,12 were selected for full text examination and subsequent inclusion in the review and a synopsis of each study is provided below.

1. Zhang et al, 2018 Authors, affiliation and funding: Zhang M. Solomon DH, Desai RJ, Kang EH, Liu J, Neogi T, Kim SC. Brigham and Women’s Hospital, Harvard Medical School, Boston, MA, no specific funding specified. Objectives: To examine the risk of CV events in gout patients with and without baseline CVD who started febuxostat or allopurinol (both at typical and equipotent dosing) using longitudinal comprehensive medical and pharmacy dispensing claims data from the U.S. Medicare Methods: A retrospective cohort study was conducted among those with a gout diagnosis in the Medicare population from January 1, 2008 to December 31 ,2013. Gout patients aged 65+ who initiated either allopurinol or febuxostat were identified and followed until drug discontinuation, last day of study database, Medicare disenrollment, occurrence of outcome, death, or nursing home admission. Initiators of febuxostat or allopurinol had no use of the initiating drug in 365 days before their first prescription. However, a febuxostat initiator may have had allopurinol in the 365 days prior and vice versa for allopurinol. New use, defined as naivety to both drugs, was examined in a sensitivity analysis. Exclusion criteria included: age younger than 65 years on the index date, less than 365 days of insurance eligibility in Medicare Parts A/B/D prior to the index date, no active claim in the 365 days prior to the index date, use of pegloticase or rasburicase in the 365 days prior to the index date, and ESRD/dialysis in the 365 days prior to the index date.

Outcomes: The primary outcome was defined as a composite endpoint of hospitalization for MI or stroke (excluding transient ischemic attacks (TIA). Secondary outcomes included: hospitalization for MI, stroke, coronary revascularization, HF subdivided into new onset HF or HF exacerbation, and all-cause mortality. Outcomes were identified with previously validated claims-based algorithms with positive predictive values over 80%.

d The term "drug uses" refers to mentions of a drug in association with a diagnosis during a patient visit to an office-based physician. This term may be duplicated by the number of diagnosis for which the drug is mentioned. It is important to note that a "drug use" does not necessarily result in a prescription being generated. Rather, the term indicates that a given drug was mentioned during an office visit.

116 of 169 Statistical analysis: Propensity score (PS) matching was used to adjust for potential confounders (demographic data, index year, cardiovascular comorbidities, gout-related medications, other medications and healthcare utilization patterns). Febuxostat initiators were matched to allopurinol initiators with a fixed ratio of 1:3. Baseline characteristics of febuxostat and allopurinol initiators were compared before and after PS matching. Variables with standardized differences <10% between the two groups were considered well-balanced after PS matching.

Primary as-treated analysis: follow-up started on the day after the index date and a 30-day extension was applied to the last prescription date. Treatment adherence was calculated using a proportion of days covered (PDC), where PDC (%) was equal to the number of days covered by prescriptions multiple by 100, divided by the total number of days of follow-up.

Secondary intention-to-treat 365-day analysis: follow-up time was truncated on the 366th day after the index date (one year after) unless patients were censored based on the previously mentioned criteria except drug discontinuation. This analysis was conducted to address the potential for lower adherence over long term follow-up.

Crude incidence rates (95% CI) were reported for the primary and secondary outcomes in the PS-matched cohorts. Cox proportional hazards regression was used to compare the risk of primary and secondary outcomes in the PS matched cohorts of febuxostat and allopurinol initiators. As the proportional hazards assumption was violated for the all-cause mortality analysis, a Cox regression stratified by follow-up time (i.e., treatment duration) for all-cause mortality was conducted. In addition, Cox regression stratified by follow-up time was conducted for the primary outcome.

Subgroup analyses: Two subgroup analyses were conducted. The first was the primary analysis stratified by presence or absence of baseline CVD (defined as history of MI, hospitalized unstable angina, coronary or cerebral revascularization, stroke, or hospitalized TIA). The second subgroup included patients with peripheral vascular disease or diabetes mellitus recorded at baseline and previously defined CVD to reflect the CARES study inclusion criteria.

Two sensitivity analyses were conducted. The first was restricted to patients who initiated equipotent dosing of febuxostat (≥40mg daily) versus allopurinol (≥300mg daily) on the index date. The second analysis was restricted to patients who initiated equipotent dosing of febuxostat (≥40mg daily) versus allopurinol (≥300mg daily) on the index date and were naïve to both drugs prior to index date.

Results: There were 331,134 gout patients identified and following 1:3 PS matching, 95% of febuxostat initiators (n=24,936) and 25% of allopurinol initiators (n=74,808) were included in the study. All baseline covariates were well-balanced between the PS-matched groups with a standardized difference <10%.

The mean [standard deviation (SD)] follow-up time was 1.1 (1.1) years among febuxostat initiators and 1.2 (1.2) years among allopurinol initiators. Of the febuxostat initiators, 30.4% had been on allopurinol at some point during the 365 days prior to febuxostat initiation. Among allopurinol initiators, 0.4% had been on febuxostat during the 365 days prior to allopurinol initiation.

The incidence rate per 100 person-years for the primary outcome (a composite of hospitalization for MI or stroke) was 3.43 (CI: 3.22-3.66) in febuxostat and 3.36 (CI: 3.25-3.49) in allopurinol initiators. The hazard ratio (HR) for the primary outcome in the as treated analysis was 1.01 (95% CI 0.94-1.08) for febuxostat compared with allopurinol initiators and when stratified by follow-up time the HR (95% CI) was 0.84 (0.73-0.98) for 0-1 year of follow-up, 0.88 (0.61-1.25) for 1-2 years, 0.76 (0.42-1.39) for 2-3 years, and 1.17 (0.45-3.05) for >3 years. For the intention to teat analysis, the HR 95% CI was 0.97(0.89- 1.06).

117 of 169 The incidence rates per 100 person-years for individual secondary outcomes including hospitalization for MI, stroke, coronary revascularization, HF and all-cause mortality were similar between febuxostat and allopurinol initiators yielding a HR of 1.03 (95% CI 0.94-1.13) for MI, 0.98 (95% CI 0.87-1.10) for stroke, 0.95 (95% CI 0.87-1.03) for coronary revascularization, 0.95 (95% CI 0.89-1.02) for all-cause mortality new-onset HF was 1.05 (95% CI 0.98-1.12) for new-onset HF and 0.94 (95% CI 0.91-0.99) for HF exacerbation in the as treated analyses. Results from the ITT analysis were similar (Table 1). In the Cox regression analysis stratified by follow-up time for all-cause mortality, the HR (95% CI) associated with febuxostat was 0.75 (0.66-0.86) for 0-1 year of follow-up, 0.85 (0.63-1.15) for 1-2 years, 0.72 (0.53- 1.54) for 2-3 years, and 1.25 (0.56-2.80) for >3 years.

In the first subgroup analysis by baseline CVD the results did not differ from the main analysis for the primary outcome. For all-cause mortality, the HR for febuxostat versus allopurinol was 0.97 (95% CI 0.90-1.04) in those without baseline CVD and 0.85 (95% CI 0.72-0.99) among those with baseline CVD (Table 2). In the second subgroup analysis based on high CV risk, which was defined like the CARES inclusion criteria, no difference was seen in both primary and secondary outcomes, including all-cause mortality (Table 3).

For the sensitivity analysis limited to patients who initiated febuxostat ≥40mg daily versus allopurinol ≥300mg daily results for all outcomes were consistent with the main analyses. When restricted to those who initiated equipotent dosing and had no prior use of either febuxostat or allopurinol before the index date, no difference in risk was seen between the two groups for the primary outcome (HR 0.96, 95% CI 0.85-1.08) as well as all secondary outcomes.

Table 1. Risk of cardiovascular events and mortality in febuxostat versus allopurinol initiators (extracted from publication).

118 of 169

Table 2. Risk of cardiovascular events and mortality in febuxostat versus allopurinol initiators by baseline CVD (extracted from publication)

Table 3. Risk of cardiovascular events and mortality in febuxostat versus allopurinol initiators by high CV risk (extracted from publication)

119 of 169

Strengths • The use of an active comparator design minimized unmeasured confounding. • Propensity score matching was performed to balance baseline covariates between both exposure groups • Large sample size • Validated algorithms were used for outcome measurement.

Limitations • Main analysis used a prevalent new user design to reflect typical prescription practices for febuxostat initiation. Although a new user design was applied in a sensitivity analysis, the analysis was restricted to those who initiated equipotent dosing of febuxostat (≥40mg daily) and allopurinol (≥300mg daily) and not the entire study cohort. • No information was available on cause-specific mortality which was examined in the CARES trial. • No information was available on family history of CVD, severity of gout, serum urate levels and use of over-the-counter medications such as NSAIDs or aspirin, which could have led to residual confounding. • Limited generalizability as cohort was restricted to those aged over 65 years • Limited follow-up time (mean follow-up time was 1.2 years for the allopurinol cohort and 1.1 years for febuxostat cohort), which led to less precise estimates for the long-term effects of febuxostat

Authors Conclusions: Overall there was no difference in the risk of myocardial infarction, stroke, coronary revascularization, new heart failure, or all-cause mortality between febuxostat and allopurinol initiators. However, a trend toward an increased risk of all-cause mortality in long-term users of febuxostat (>3 years) was observed. The risk of HF exacerbation was slightly lower among febuxostat compared to allopurinol initiators.

120 of 169

2. Foody et al, 2017

Authors, affiliation and funding: Foody J, Turpin RS, Tidwell BA, Lawrence D, Schulman KL. Harvard Medical School, Boston, MA, and, Janssen Pharmaceuticals, New Brunswick, NJ. Takeda Pharmaceuticals USA, Deerfield, IL, Outcomes Research Solutions, Shrewsbury, MA. This study was funded by Takeda Pharmaceuticals who are the sponsor of febuxostat.

Objectives: The goal of this study was to evaluate the impact of initiating allopurinol or febuxostat on major CV events in patients with gout, pre-existing CVD or HF, and stage 3 or 4 CKD in a real-world setting.

Methods: A retrospective cohort study was conducted in the Truven MarketScan Commercial Claims and Encounters and Medicare Supplemental and Coordination of Benefits databases. Patients aged ≥18 years or more who had a diagnosis of gout, stage 3 or 4 CKD, and a history of CVD or HF and who received first-line treatment with allopurinol or febuxostat for ≥31 days of continuous therapy between January 1, 2009 and June 30, 2013 were identified. Gout was defined as ≥1 diagnosis of gout (ICD-9-CM code 274.xx) and CKD was defined having (ICD-9-CM codes 585.3 or 585.4) on a claim during the 12-month baseline period after (assume typo and intended to say before) the study index date which was the date of initiating allopurinol or febuxostat. A history of CVD was defined as having an ICD-9-CM diagnosis of coronary artery disease (CAD), cerebrovascular disease, or peripheral vascular disease (PVD) on ≥1 inpatient claims or ≥2 non-diagnostic outpatient claims, or evidence of a previous revascularization procedure or non-traumatic lower-extremity amputation. Only those who were continuously enrolled in the MarketScan databases for ≥12 months before (baseline) and ≥31days after the index date, with complete data availability were included. Patients were excluded if they received pegloticase ≤2 months before the index date, or if they had evidence of organ transplant at baseline, ESRD, a non-skin malignancy, or HIV/AIDS. Patients whose baseline CV or HF status could not be determined were also excluded from the study. Patients were followed from index date until disenrollment from the MarketScan databases, discontinuation of the qualifying study drug, use of the alternate study drug, onset of any of the exclusion disease states, or post-index exposure to pegloticase.

Outcomes: The primary outcome was the occurrence of major CV events including CAD-specific, cerebrovascular disease–specific, and PVD-specific events. The secondary outcome was incidence of gout flares, defined as (1) an office visit for gout with the diagnosis of gout and a new dispensing of colchicine, selective or nonselective NSAIDs, or oral or injectable glucocorticoids within 14 days of the office visit or (2) an emergency department or inpatient visit coded for gout.

Statistical analysis: Cox proportional hazards models assessed the predictors of any major CV event in aggregate, and of CAD, cerebrovascular disease, and PVD events individually. The Cox proportional hazards model was tested to ensure that proportionality assumptions were met. Appropriate fit was assessed using the likelihood ratio, score, and Wald tests. Explanatory variables tested for inclusion in the model included basic demographics, baseline CVD type, baseline history of other comorbid conditions, baseline medication exposure, and CKD stage nearest the index date. For the secondary outcome, an additional extended Cox model (Anderson-Gill) was developed to assess the relative hazard of the repeated measure of gout flare.

Results: There were 1.2 million adults with a diagnosis of gout in the MarketScan databases between January 1, 2009, and June 30, 2013, and 14,998 patients who initiated treatment with allopurinol or with febuxostat and had a diagnosis of stage 3 or 4 CKD. After exclusions the study included 370 febuxostat and 2056 allopurinol initiators. Of note, the largest exclusion was due to enrollment and inclusion criteria (n=6845). The mean duration of follow-up was 9 months (standard deviation [SD], 7.5 months) in the febuxostat cohort and 9.2 months (SD, 8.2 months) in the allopurinol cohort. The median daily dose was

121 of 169 150 mg (interquartile range [IQR], 100-250 mg) in the allopurinol cohort and 40 mg (IQR, 40-60 mg) in the febuxostat cohort.

Allopurinol users were more likely than febuxostat users to have had HF (51.8 vs 44.3 p=0.09) and COPD (21.7 vs 16.2 p=0.016) at baseline, and febuxostat users were more likely to have had aneurysms (1.9 vs 0.9 p=0.096). Febuxostat users were more likely than allopurinol users to use colchicine (49.2 vs 35.6 p < 0.001), glucocorticoids (63.5 vs 57.6 p=0.033) and probenecid (4.3 vs 1.3 p=0.001). They were also more likely to have a higher mean number of medication classes (8.3 vs 7.9 p=0.001).

Major CV events Overall, 162 (6.7%) patients had at least 1 of the major CV events during follow-up (3.8% among febuxostat users vs 7.2% among allopurinol users; P = .015) with an unadjusted incidence rate per 1000 person-years of 51.8 (95% CI, 28-87) in the febuxostat cohort and 99.3 (95% CI, 84-117) in the allopurinol cohort. There was an increased likelihood of any major CV event among patients with PVD (hazard ratio [HR], 2.69; 95% CI, 1.95-3.71) or a baseline CVD or HF hospitalization (HR,1.75; 95% CI, 1.22-2.50;), and a decreased likelihood of any major CV event among febuxostat initiators (HR, 0.52; 95% CI, 0.30-0.91; patients with a baseline history of an “other” circulatory disorder; HR, 0.54; 95% CI, 0.35-0.85;), and those receiving baseline antigout medications, (HR, 0.69; 95% CI, 0.48-1.00) (Appendix Figure 1)

CAD-specific major CV event The incidence rates per 1000 person- years for CAD-specific major CV event were 29.4 (95% CI, 13-58) and 46.8 (95% CI,37-59) in the febuxostat and allopurinol cohorts, respectively. There was a significantly increased likelihood of CAD major CV events among patients with baseline PVD (HR, 2.04; 95% CI, 1.27-3.26) and baseline CVD or HF hospitalization (HR, 1.83; 95% CI, 1.11-3.02).

Cerebrovascular disease–specific major CV event The incidence rate per 1000 person-years for cerebrovascular disease–specific major CV event was 18.2 (95% CI, 6-43) for febuxostat and 21.4 (95% CI, 15-30) for allopurinol users. The likelihood of a cerebrovascular-specific major CV event was increased among patients with baseline cerebrovascular disease and those with a baseline CVD or HF hospitalization. In addition, a significantly decreased likelihood of cerebrovascular disease–related major CV events was seen in female patients and those with a history of a non-CVD circulatory system disorder.

PVD-specific major CV event The incidence rates per 1000 person-years of a PVD-specific major CV event were 7.3 (95% CI, 1-26) and 31.8 (95% CI, 24-42) in the febuxostat and allopurinol cohorts, respectively. The likelihood of a PVD-related major CV event was significantly increased among patients with a baseline PVD event and among those aged 55 to 74 years at the time of treatment initiation compared with being aged ≥75 years. Febuxostat initiation was associated with a significant reduction in PVD-related major CV events.

Strengths • The use of an active comparator design minimized unmeasured confounding. • Validated algorithms were used for some outcome measurements.

Limitations • No primary statistical hypothesis was specified, and multiple predictors of the outcomes were evaluated in cox models. • There was no indication of when follow up started. From the description in the methods section it could have started at index date or 31 days after index date. If it was 31 days after the index date, this may have caused under ascertainment of the outcome by missing early events. • There was no indication if ULT users were new users or prevalent users.

122 of 169 • Minimal adjustment for confounding in a multivariate cox model • There was no indication of when CVD history was determined. • Duration of follow-up was very short: 9 months (standard deviation [SD], 7.5 months) in the febuxostat cohort and 9.2 months (SD, 8.2 months) in the allopurinol cohort. • No information was available on cause-specific mortality which was examined in the CARES trial. • No information was available on family history of CVD, serum urate levels and use of over-the- counter medications such as NSAIDs or aspirin, which could have led to residual confounding.

Authors Conclusions: The study results suggest that patients with gout and moderate-to-severe CKD and CVD or HF who initiate febuxostat treatment may have lower rates of major CV events than patients who initiate allopurinol therapy

3.2 DRUG UTILIZATION RESULTS

3.2.1 Sales Distribution Data

3.2.1.1 Settings of Care Manufacturer sales data for 2017 indicated that approximately 77% of febuxostat, 78% of allopurinol, and 76% of probenecid products were primarily sold to U.S. outpatient retail pharmacies.e Approximately 67% of IV rasburicase and 81% of IV pegloticase products were sold to non-retail settings (primarily non- federal hospitals and clinics). Given that the majority of febuxostat and allopurinol sales were to U.S. outpatient retail pharmacies, our analysis focused on outpatient utilization data. Drug utilization data from mail-order/specialty pharmacies and non-retail settings of care, such as hospitals and clinics, were not included in this analysis.

3.2.2 U.S. Outpatient Retail Pharmacy Data

3.2.2.1 Prescription-Level Data Figure 1 below and Table 1 in Section 8.2 provides the nationally estimated number of prescriptions dispensed for febuxostat, allopurinol, or probenecid from U.S. outpatient retail pharmacies from 2013 through 2017. An estimated 16.2 million total prescriptions were dispensed for these ULTs in 2017, of which allopurinol accounted for the majority of use at 92% (14.9 million prescriptions) total prescriptions. Febuxostat and probenecid accounted for 7% and 1% (1.2 million and 152,000 prescriptions) of total prescriptions, respectively. Over the last five years, prescription utilization for allopurinol increased 8% from approximately 13.8 million prescriptions in 2013 to 14.9 million prescriptions in 2017. Prescriptions dispensed for febuxostat also increased 18% from approximately 1 million prescriptions in 2013 to 1.2 million prescriptions in 2017. Meanwhile, the utilization of probenecid decreased 27% from 208,000 prescriptions in 2013 to 152,000 prescriptions in 2017 (Figure 1). Figure 1. Nationally estimated number of prescriptions dispensed for febuxostat, allopurinol, or probenecid from U.S. outpatient retail pharmacies, 2013 through 2017

e IQVIA™ National Sales Perspective. 2017. Extracted October 2018. File: NSP 2018-1544 febuxostat and other molecules by Sup Ch 10-30-2018.

123 of 169

Analysis of U.S. outpatient retail pharmacy data showed that allopurinol was the most widely used ULT in 2017, with 14.9 million prescriptions dispensed compared to 1.2 million prescriptions for febuxostat. The nationally estimated prescription data suggests that use of allopurinol and febuxostat increased between 2013-2017. The most commonly dispensed strengths were allopurinol 300 mg tablets and febuxostat 40 mg tablets. Utilization by sex and age showed that male patients and patients aged 65 years or older accounted for the largest proportion of allopurinol and febuxostat use. Of note, data on indication for use directly linked to dispensed prescription data are not available; however, diagnoses and dosing information were obtained from a monthly survey of 3,200 office-based physicians, representing 30 specialties, across the U.S. These physicians report all patient interactions during one typical workday a month, which are then projected to provide national estimates. According to this survey data, gout was the top diagnosis linked to discussion of allopurinol and febuxostat. Dosing information showed that febuxostat 40 mg or 80 mg tablets prescribed once a day, or allopurinol 100 mg or 300 mg tablets prescribed once a day, were the most common directions of use reported for these XOIs, for gout, which is consistent with the labeled indication for both products. The physician survey data is limited in that it does not provide a direct measure of the number of patients or prescriptions but was included in this review to describe typical ULT prescribing in office-based physician’s clinical practice. Therefore, the sample of office-based physicians included in the survey may not represent prescribing practices for ULTs in other specialties or settings of care. Further limitations are summarized in section 8.1. Two studies were identified for inclusion in the literature review which aimed to summarize the findings from observational studies on the risk of CV events and mortality in febuxostat compared to allopurinol users, in real world settings. Zhang et al, found that among U.S. Medicare beneficiaries with gout, initiation of febuxostat compared with allopurinol was not associated with an increased risk of CV events including, MI, stroke, and coronary revascularization, which is in line with the findings from the CARES study. However, unlike the CARES study, Zhang et al reported no difference in risk for all-cause mortality in both the main analysis and in subgroup analyses restricted to those with baseline CVD and high CVD risk. These results were observed in the as-treated as well as intention-to-treat analyses. Despite this, when the main analysis was stratified by follow-up time, a slightly higher risk of all-cause mortality was seen in febuxostat users with more than 3 years of follow-up. A modestly decreased risk for HF exacerbation associated with febuxostat versus allopurinol use was also seen. Foody et al, reported that among patients with gout and moderate-to-severe CKD and CVD or HF, initiation of febuxostat was associated with a lower risk of major CV events. However, the methods applied in the Foody study were limited by a number of issues including: limited adjustment for important confounders, unclear reporting of a number of important elements such as, whether ULT users were new or prevalent users, when history of CVD was determined and when follow up started, and very short follow up times in both the allopurinol and febuxostat cohorts, Additionally, no clear primary statistical hypothesis was specified and multiple predictors of the outcomes were evaluated in the Cox proportional hazards models, in the statistical analysis. Given these limitations, and the fact that the results of this study were inconsistent with the results of both the CARES study and the study by Zhang et al, it was challenging to meaningfully interpret the results for this review. Based on the this, the remainder of this discussion will focus on the study by Zhang et al, which had fewer limitations and was reported with more clarity.

The CARES study examined both CV and all-cause mortality and found an increased risk of both; however, Zhang et al were limited to examining all-cause mortality, as cause of death was not available in the Medicare database. They found overall no increased risk of all-cause mortality, but when stratifying by follow-up time (treatment duration) there was a suggested non-significant increased risk of all-cause mortality in longer-term (>3 years) febuxostat users compared to allopurinol users HR 1.25 (0.56-2.80). The number of patients in this analysis were not shown in the paper, but the precision of these estimates may have been affected by small numbers of long term ULT users in this study, as the mean follow-up time was just 1.2 years. Adherence to ULT decreased over time in this study and given the short study period (2009-2013) there may have been low numbers of longer term febuxostat users. The CARES study enrolled only gout patients with a history of major CVD including MI, hospitalization for unstable angina, stroke, hospitalization for transient ischemic attack, peripheral vascular disease, or

126 of 169 diabetes mellitus with evidence of microvascular or macrovascular disease. However, Zhang et al included all gout patients in their cohort irrespective of CVD history and did two sub-group analyses: one which stratified the main analysis by baseline CVD and one which stratified the main analysis by high CV risk at baseline according to the CARES trial inclusion criteria (above). The first subgroup analysis for all-cause mortality reported a HR for febuxostat versus allopurinol of 0.97 (95% CI 0.90-1.04) in those without baseline CVD and 0.85 (95% CI 0.72-0.99) among those with baseline CVD. However, in the second subgroup no mortality difference was seen in those with high CV risk at baseline and those without.

The study by Zhang et al was well conducted using an active comparator to reduce unmeasured confounding, propensity score matching to balance comparator cohorts on known potential confounders, validated algorithms for outcome ascertainment and both as treated and ITT type analyses to reflect risks among various exposure patterns. However, a prevalent new user design was applied in the main analysis and sub-group analyses whereby new febuxostat users may have used allopurinol prior to initiating febuxostat and new allopurinol users may have used febuxostat prior to initiating allopurinol. Prevalent new user designs can lead to issues with selection bias, for example; prevalent new febuxostat users were survivors of the earlier allopurinol use period. Immortal time bias may also be an issue as the time on allopurinol prior to febuxostat initiation is immortal and is not accounted for in the study. Prevalent new user designs also raise concerns about potentially adjusting for intermediate risk factors on the causal pathway, as covariates, determined in the baseline period, may have been influenced by use of the comparator drug in that period. Despite this, Zhang et al did conduct a sensitivity analysis in which users were naive to both drugs in the 365 days prior to initiation, reflecting a new user design. However, this analysis was restricted to those who initiated equipotent dosing of febuxostat (≥40mg daily) and allopurinol (≥300mg daily) and not the entire study cohort. The results of this new user sensitivity analysis were not substantially different from the main prevalent new user analysis. The study was also limited in that it did not include information on OTC NSAIDs and aspirin which could have led to residual confounding, as NSAID and aspirin use can influence CV events and NSAIDS are used commonly in gout patients. Febuxostat users may have more severe gout given it is often used in practice as a second line agent compared to allopurinol and many commercial insurance plans require that patients fail allopurinol treatment to justify the inflated cost of febuxostat. If this were the case febuxostat users may be more likely to use NSAIDs to manage more severe gout symptoms and NSAIDs can also increase the risk of CV events.

The study was large including 24,936 febuxostat initiators and 74,808 allopurinol initiators in the prevalent new user main analysis and 11,220 and 33660 initiators of febuxostat 40mg and allopurinol 300mg, respectively, in the new user sensitivity analysis. The Medicare study population was older and more balanced in gender compared to the CARES trial (52% male with a median age of 76 years in Medicare versus 84% male with a median age of 64 years in CARES). However, the older age also limits the generalizability of the findings to those aged < 65 years.

While the findings of this observational study agree with the CARES study, in relation to the null findings for risk of CV events, it is not clear why the null findings for all-cause mortality, especially in the subgroup analyses by both baseline CVD and high-risk CVD, differed from those of CARES, where an increased risk of all-cause mortality was seen. One potential reason for this may be differences in the populations in both studies. In the study by Zhang et al, participants tended to be older than those enrolled in the CARES study and only 12 % were recorded as having baseline CVD. Therefore, numbers of febuxostat and allopurinol users in the subgroup analyses by CVD were small. Inability to properly determine CVD history based on the relatively short baseline look back period in the claims data may have influenced these findings as one would expect prevalence of CVD in older patients to be higher than 12%.

5 CONCLUSIONS

127 of 169 Use of ULTs appears to be increasing over time in the US and the most commonly prescribed ULT is allopurinol. Based on the findings of the study by Zhang et al, there is no evidence that in gout patients febuxostat use increases CV events or all-cause mortality compared to allopurinol use, in real world settings. There was some suggestion of an increased risk of mortality in longer term users; however, numbers were likely small, and larger studies with more longer-term users are needed to confirm this finding.

6 RECOMMENDATIONS DEPI has no recommendations at this time. The need for further CV safety data from long-term users of febuxostat and allopurinol should be addressed in the FAST trial, currently underway in the UK and Denmark.

7 REFERENCES

128 of 169 1. Schumacher H CL. Gout and other crystal-associated arthropathies. In: Kasper D FA, Hauser S, Longo, D JJ, Loscalzo J. , eds. Harrison's Principles of Internal Medicine. 19th ed. New York: McGraw-Hill; 2014. 2. Hui M, Carr A, Cameron S, et al. The British Society for Rheumatology Guideline for the Management of Gout. Rheumatology (Oxford) 2017;56:1246. 3. Khanna D, Fitzgerald JD, Khanna PP, et al. 2012 American College of Rheumatology guidelines for management of gout. Part 1: systematic nonpharmacologic and pharmacologic therapeutic approaches to hyperuricemia. Arthritis Care Res (Hoboken) 2012;64:1431-46. 4. Richette P, Doherty M, Pascual E, et al. 2016 updated EULAR evidence-based recommendations for the management of gout. Ann Rheum Dis 2017;76:29-42. 5. Sivera F, Andres M, Carmona L, et al. Multinational evidence-based recommendations for the diagnosis and management of gout: integrating systematic literature review and expert opinion of a broad panel of rheumatologists in the 3e initiative. Ann Rheum Dis 2014;73:328-35. 6. Grimaldi-Bensouda L, Alperovitch A, Aubrun E, et al. Impact of allopurinol on risk of myocardial infarction. Ann Rheum Dis 2015;74:836-42. 7. Okafor ON, Farrington K, Gorog DA. Allopurinol as a therapeutic option in cardiovascular disease. Pharmacol Ther 2017;172:139-50. 8. Richette P, Perez-Ruiz F, Doherty M, et al. Improving cardiovascular and renal outcomes in gout: what should we target? Nat Rev Rheumatol 2014;10:654-61. 9. MacDonald TM, Ford I, Nuki G, et al. Protocol of the Febuxostat versus Allopurinol Streamlined Trial (FAST): a large prospective, randomised, open, blinded endpoint study comparing the cardiovascular safety of allopurinol and febuxostat in the management of symptomatic hyperuricaemia. BMJ Open 2014;4:e005354. 10. Kim SC, Schneeweiss S, Choudhry N, Liu J, Glynn RJ, Solomon DH. Effects of xanthine oxidase inhibitors on cardiovascular disease in patients with gout: a cohort study. Am J Med 2015;128:653 e7- e16. 11. Zhang M, Solomon DH, Desai RJ, et al. Assessment of Cardiovascular Risk in Older Patients with Gout Initiating Febuxostat versus Allopurinol: A Population-Based Cohort Study. Circulation 2018. 12. Foody J, Turpin RS, Tidwell BA, Lawrence D, Schulman KL. Major Cardiovascular Events in Patients with Gout and Associated Cardiovascular Disease or Heart Failure and Chronic Kidney Disease Initiating a Xanthine Oxidase Inhibitor. Am Health Drug Benefits 2017;10:393-401.

8 APPENDICES

129 of 169 8.1 DRUG UTILIZATION DATABASE DESCRIPTIONS AND LIMITATIONS IQVIA™ National Sales Perspectives (NSP) The IQVIA™ National Sales Perspectives™ (NSP) measures the volume of drug products, both prescription and over-the-counter, and selected diagnostic products moving from manufacturers into various outlets within the retail and non-retail markets. Volume is expressed in terms of sales dollars, extended units, and share of market. These data are based on national projections. Outlets within the retail market include the following pharmacy settings: chain drug stores, independent drug stores, mass merchandisers, food stores, and mail service. Outlets within the non-retail market include clinics, non- federal hospitals, federal facilities, HMOs, long-term care facilities, home health care, and other miscellaneous settings.

IQVIA™ National Prescription Audit (NPA) The IQVIA™ National Prescription Audit (NPA) measures the “retail outflow” of prescriptions, or the rate at which drugs move out of retail pharmacies, mail service houses, or long-term care facilities into the hands of consumers via formal prescriptions in the U.S. The NPA audit measures what is dispensed by the pharmacist. Data for the NPA audit is a national level estimate of the drug activity from retail pharmacies. NPA receives over 3.7 billion prescription claims per year, captured from a sample of the universe of approximately 59,900 pharmacies throughout the U.S. The pharmacies in the database account for most retail pharmacies and represent nearly 93% of retail prescriptions dispensed nationwide. The type of pharmacies in the sample are a mix of independent, retail, chain, mass merchandisers, and food stores with pharmacies, and include prescriptions from cash, Medicaid, commercial third-party and Medicare Part-D prescriptions. Data is also collected from approximately 45 – 75% (varies by class and geography) of mail service pharmacies and approximately 71 – 83% of long-term care pharmacies. Data are available on-line for 72-rolling months with a lag of 1 month.

IQVIA™ Total Patient Tracker (TPT) The IQVIA™ Total Patient Tracker (TPT) is a national-level projected service designed to estimate the total number of unique (non-duplicated) patients across all drugs and therapeutic classes in the retail outpatient setting from U.S. retail pharmacies. Data are available back to January 2002 and are available 20 days after the close of the month. TPT uses prescription activity as part of its projection and integrates information from pharmacies and payers to eliminate duplicate patients and multiple prescription fills, producing quick and reliable unique patient counts. IQVIA has 93% coverage and a sample of ~59,900 retail pharmacies. IQVIA captures about 3.8 billion transactions annually. TPT is projected to the known universe of retail pharmacies.

Syneos Health Research & Insights TreatmentAnswers™ The Syneos Health Research & Insights TreatmentAnswers™ and TreatmentAnswers™ with Pain Panel is a monthly survey designed to provide descriptive information on the patterns and treatment of diseases encountered in office-based physician practices in the U.S. The survey consists of data collected from over 3,200 office-based physicians representing 30 specialties across the United States that report on all patient activity during one typical workday per month. These data may include profiles and trends of diagnoses, patients, drug products mentioned during the office visit and treatment patterns. The Pain Panel supplement surveys over 115 pain specialist physicians each month. With the inclusion of visits to pain specialists, this will allow additional insight into the pain market. The data are then projected nationally by physician specialty and region to reflect national prescribing patterns.

130 of 169 The office-based physician survey data was used to provide insight into prescriber intent for the typical uses for a drug in an office-based physician’s clinical practice. Thus, the sample of office-based physicians might not well represent the care from specialty offices or inpatient settings where prescribing for the treatment of hyperuricemia in patients with gout may have occurred. These data are nationally estimated and due to the small sample sizes, less than 100,000 occurrences do not represent a sufficient portion of the population and is not representative of total physician prescribing habits at a national level. It’s important to note that drug use mentions do not necessarily result in a prescription being generated but are the projected number of times a drug was mentioned during an office visit.

Limitations The overall findings from this analysis should be interpreted in the context of the known limitations of the databases used. Because we focused our analysis on outpatient retail pharmacy settings, the patient exposure estimates reported in this review can only be generalized to these settings of care and may not apply to other settings in which these products may be prescribed or dispensed such as mail- order/specialty pharmacies or hospitals and various other clinical settings where patients receive health care. In addition, information, such as patient age, cannot be validated due to lack of access to medical records. Therefore, our results may underestimate patient utilization of ULTs. Furthermore, the estimates provided are national estimates, no statistical tests were performed to determine statistical significant changes over time or between products. Therefore, all changes over time or between products should be considered approximate.

131 of 169 8.2 DRUG UTILIZATION TABLES Table 1. Nationally estimated number of prescriptions dispensed for febuxostat, allopurinol, or probenecid from U.S. outpatient retail pharmacies 2013 through 2017 2013 2014 2015 2016 2017 TRx Share TRx Share TRx Share TRx Share TRx Share N % N % N % N % N % TOTAL PRESCRIPTIONS 14,981,040 100.0% 15,509,230 100.0% 15,835,819 100.0% 16,234,046 100.0% 16,241,079 100.0% allopurinol 13,774,089 91.9% 14,252,919 91.9% 14,545,593 91.9% 14,900,713 91.8% 14,914,431 91.8% febuxostat 999,193 6.7% 1,066,587 6.9% 1,115,912 7.0% 1,167,587 7.2% 1,174,842 7.2% probenecid 207,758 1.4% 189,724 1.2% 174,314 1.1% 165,746 1.0% 151,806 0.9% Data Source: IQVIA™ National Prescription Audit. Years 2013-2017. Extracted October 2018. File: NPA febuxostat, allopurinol & probenecid Retail TRxs 9-28-2018

Table 2. Nationally estimated number of prescriptions dispensed for febuxostat or allopurinol from U.S. outpatient retail pharmacies, stratified by drug strength, 2013 through 2017 2013 2014 2015 2016 2017 TRx Share TRx Share TRx Share TRx Share TRx Share N % N % N % N % N % TOTAL PRESCRIPTIONS 14,773,282 100.0% 15,319,506 100.0% 15,661,505 100.0% 16,068,300 100.0% 16,089,273 100.0% allopurinol 13,774,089 93.2% 14,252,919 93.0% 14,545,593 92.9% 14,900,713 92.7% 14,914,431 92.7% 300 mg strength tablets 7,594,356 55.1% 7,746,152 54.3% 7,800,890 53.6% 7,874,320 52.8% 7,744,761 51.9% 100 mg strength tablets 6,179,733 44.9% 6,506,767 45.7% 6,744,702 46.4% 7,026,393 47.2% 7,169,670 48.1% febuxostat 999,193 6.8% 1,066,587 7.0% 1,115,912 7.1% 1,167,587 7.3% 1,174,842 7.3% 40 mg strength tablets 687,216 68.8% 732,520 68.7% 764,664 68.5% 798,627 68.4% 801,453 68.2% 80 mg strength tablets 311,977 31.2% 334,067 31.3% 351,248 31.5% 368,960 31.6% 373,389 31.8% Data Source: IQVIA™ National Prescription Audit. Years 2013-2017. Extracted October 2018. File: NPA febuxostat & allopurinol by strength Retail TRxs 10-30-2018 Note: allopurinol 500 mg strength (IV injectable) was not included in the table above due to the negligible amount of use observed in the outpatient retail setting

132 of 169 Table 3. Nationally estimated number of patients who received a dispensed prescription for febuxostat or allopurinol from U.S. outpatient retail pharmacies, stratified by patient sex and age groups, 2013 through 2017 2013 2014 2015 2016 2017 Patients Share Patients Share Patients Share Patients Share Patients Share N % N % N% N%N % TOTAL PATIENTS* 2,927,702 100.0% 3,025,303 100.0% 3,072,783 100.0% 3,195,168 100.0% 3,298,703 100.0% allopurinol 2,765,083 94.4% 2,854,350 94.3% 2,895,059 94.2% 3,008,553 94.2% 3,105,778 94.2% male patients 2,018,956 73.0% 2,097,807 73.5% 2,155,187 74.4% 2,217,855 73.7% 2,276,512 73.3% ≤ 20 years 2,165 0.1% 4,865 0.2% 3,574 0.2% 3,660 0.2% 3,131 0.1% 21-44 years 222,321 11.0% 232,653 11.1% 239,689 11.1% 237,999 10.7% 236,784 10.4% 45-64 years 860,547 42.6% 897,578 42.8% 937,094 43.5% 952,421 42.9% 962,944 42.3% 65 years or older 920,891 45.6% 967,455 46.1% 1,019,928 47.3% 1,059,116 47.8% 1,104,499 48.5% unspecified age 140,645 7.0% 83,609 4.0% 17,925 0.8% 1,687 0.1% 4,244 0.2% female patients 750,310 27.1% 809,853 28.4% 833,538 28.8% 841,631 28.0% 855,966 27.6% ≤ 20 years 1,124 0.1% 3,005 0.4% 2,034 0.2% 2,132 0.3% 1,800 0.2% 21-44 years 26,865 3.6% 37,522 4.6% 37,421 4.5% 33,599 4.0% 31,808 3.7% 45-64 years 204,562 27.3% 229,112 28.3% 238,956 28.7% 234,333 27.8% 233,388 27.3% 65 years or older 512,693 68.3% 542,844 67.0% 567,912 68.1% 583,384 69.3% 599,168 70.0% unspecified age 42,352 5.6% 24,444 3.0% 4,953 0.6% 1,047 0.1% 1,852 0.2% unspecified sex 13,925 0.5% 22,685 0.8% 8,771 0.3% 2,939 0.1% 2,876 0.1% febuxostat 206,831 7.1% 215,939 7.1% 223,374 7.3% 233,095 7.3% 237,618 7.2% male patients 146,746 71.0% 154,355 71.5% 162,444 72.7% 167,226 71.7% 168,698 71.0% ≤ 20 years 67 0.0% 139 0.1% 88 0.1% 119 0.1% 124 0.1% 21-44 years 21,098 14.4% 21,814 14.1% 22,766 14.0% 21,892 13.1% 20,368 12.1% 45-64 years 64,407 43.9% 68,308 44.3% 73,884 45.5% 75,176 45.0% 74,704 44.3% 65 years or older 59,754 40.7% 64,107 41.5% 68,855 42.4% 72,279 43.2% 75,356 44.7% unspecified age 9,015 6.1% 5,734 3.7% 1,126 0.7% 118 0.1% 291 0.2% female patients 60,666 29.3% 66,370 30.7% 68,746 30.8% 70,301 30.2% 71,211 30.0% ≤ 20 years 22 0.0% 45 0.1% 28 0.0% 42 0.1% 36 0.0% 21-44 years 1,785 2.9% 2,505 3.8% 2,539 3.7% 2,342 3.3% 2,062 2.9% 45-64 years 15,741 25.9% 17,839 26.9% 18,686 27.2% 18,381 26.1% 17,831 25.0% 65 years or older 42,427 69.9% 45,786 69.0% 48,265 70.2% 50,313 71.6% 51,968 73.0% unspecified age 3,295 5.4% 1,980 3.0% 422 0.6% 96 0.1% 172 0.2% unspecified sex 878 0.4% 1,657 0.8% 512 0.2% 195 0.1% 235 0.1% Data Source: IQVIA™ Total Patient Tracker. Years 2013-2017. Extracted October 2018. File: TPT Total febuxostat & allopurinol by age & gender 10-29-2018, TPT febuxostat by age & gender 10-29- 2018, TPT allopurinol by age & gender 10-29-2018 *Unique patient counts may not sum exactly due to patients aging or patients who may have received multiple products during the study period. Patients may be counted more than once in the individual age categories or across products. Therefore, summing across age groups or products is not advisable and will result in overestimates of patient counts.

133 of 169 Table 4. Nationally estimated number of prescriptions dispensed for febuxostat or allopurinol from U.S. outpatient retail pharmacies, stratified by the top 5 prescriber specialties, 2017 2017 TRx Share N % TOTAL PRESCRIPTIONS 16,089,273 100.0% allopurinol 14,914,431 92.7% Family Medicine/General Practice/Internal Medicine 8,528,450 57.2% Nurse Practitioner/Physician Assistant 2,109,521 14.1% Osteopathic Medicine 1,517,884 10.2% Nephrology 645,019 4.3% Rheumatology 621,693 4.2% All Other Specialties 1,491,864 10.0% febuxostat 1,174,842 7.3% Family Medicine/General Practice/Internal Medicine 578,214 49.2% Rheumatology 156,079 13.3% Nurse Practitioner/Physician Assistant 144,041 12.3% Osteopathic Medicine 113,552 9.7% Nephrology 100,729 8.6% All Other Specialties 82,227 7.0% Data Source: IQVIA™ National Prescription Audit. Year 2017. Extracted October 2018. File: NPA febuxostat & allopurinol by Specialty Retail TRxs 10-30-2018

134 of 169 Table 5. Top five diagnoses associated with the use of febuxostat or allopurinol as reported by U.S. office-based physician surveys, stratified by total daily dose, 2017 2017 Uses Share N % TOTAL USES 5,342,000 100.0% allopurinol 3,942,000 73.8% M10 Gout 3,437,000 87.2% 100 mg daily 1,668,000 48.6% 300 mg daily 1,321,000 38.4% 200 mg daily 205,000 6.0% 400 mg daily 42,000 1.2% Unspecified Daily Dose 166,000 4.8% All Others 34,000 1.0% E79 Disorders of purine and pyrimidine metabolism 259,000 6.6% 300 mg daily 164,000 63.3% 100 mg daily 64,000 24.7% 200 mg daily 31,000 11.9% N20 Calculus of kidney and ureter 140,000 3.6% 300 mg daily 107,000 76.2% 100 mg daily 17,000 11.9% 200 mg daily 17,000 11.9% N12 Tubulo-interstitial nephritis, not spcf as acute or chronic 35,000 0.9% 200 mg daily 35,000 100.0% N18 Chronic kidney disease (CKD) 31,000 0.8% 100 mg daily 31,000 100.0% All Other Diagnoses 41,000 1.0% febuxostat 1,400,000 26.2% M10 Gout 1,294,000 92.5% 40 mg daily 863,000 66.7% 80 mg daily 313,000 24.2% Unspecified Daily Dose 119,000 9.2% M1A Chronic gout 56,000 4.0% 80 mg daily 56,000 100.0% E79 Disorders of purine and pyrimidine metabolism 50,000 3.6% 40 mg daily 30,000 60.4% Unspecified Daily Dose 20,000 39.6% Data Source: Syneos Health Research & Insights Treatment Answers™. 2017. Extracted October 2018. File: Syneos febuxostat & allopurinol by dx & daily dose 10-30-2018 Note: Syneos Health Research & Insights Treatment Answers™ recommends caution interpreting projected annual uses below 100,000, as the sample size is very small with correspondingly large confidence intervals.

135 of 169 Table 6. Top five diagnoses associated with the use of febuxostat or allopurinol as reported by U.S. office-based physician surveys, stratified by drug strength and directions of use (signa), 2017 2017 Uses Share 95% N % Confidence Interval TOTAL USES 5,342,000 100.0% 4,910,000 - 5,774,000 allopurinol 3,942,000 73.8% 3,571,000 - 4,313,000 M10 Gout 3,437,000 87.2% 3,090,000 - 3,783,000 100 mg strength 1,933,000 56.3% 1,673,000 - 2,193,000 once a day (QD) 1,668,000 86.3% 1,427,000 - 1,910,000 twice a day (BID) 123,000 6.4% 58,000 - 189,000 2 per day 81,000 4.2% 28,000 - 135,000 four times a day (QID) 42,000 2.2% 4,000 - 81,000 less than 1 per day 11,000 0.6% <500 - 31,000 All Others 6,000 0.3% <500 - 22,000 300 mg strength 1,367,000 39.8% 1,148,000 - 1,585,000 once a day (QD) 1,314,000 96.2% 1,100,000 - 1,529,000 four times a day (QID) 16,000 1.1% <500 - 39,000 2 per day 7,000 0.5% <500 - 23,000 Unspecified signa 30,000 2.2% <500 - 62,000 Unspecified drug strength 137,000 4.0% 68,000 - 206,000 E79 Disorders of purine and pyrimidine metabolism 259,000 6.6% 164,000 - 355,000 300 mg strength 164,000 63.3% 89,000 - 240,000 once a day (QD) 164,000 100.0% 89,000 - 240,000 100 mg strength 95,000 36.7% 37,000 - 153,000 once a day (QD) 64,000 67.4% 17,000 - 111,000 2 per day 18,000 19.4% <500 - 44,000 twice a day (BID) 13,000 13.2% <500 - 34,000 N20 Calculus of kidney and ureter 140,000 3.6% 70,000 - 210,000 300 mg strength 107,000 76.2% 46,000 - 168,000 once a day (QD) 107,000 100.0% 46,000 - 168,000 100 mg strength 33,000 23.8% <500 - 67,000 once a day (QD) 17,000 50.1% <500 - 41,000 twice a day (BID) 17,000 49.9% <500 - 41,000 N12 Tubulo-interstitial nephritis, not spcf as acute or chronic 35,000 0.9% <500 - 69,000 100 mg strength 35,000 100.0% <500 - 69,000 twice a day (BID) 35,000 100.0% <500 - 69,000 N18 Chronic kidney disease (CKD) 31,000 0.8% <500 - 63,000 100 mg strength 31,000 100.0% <500 - 63,000 once a day (QD) 31,000 100.0% <500 - 63,000 All Other Diagnoses 41,000 1.0% 3,000 - 79,000 febuxostat 1,400,000 26.2% 1,179,000 - 1,621,000 M10 Gout 1,294,000 92.5% 1,081,000 - 1,507,000 40 mg strength 896,000 69.3% 720,000 - 1,073,000 once a day (QD) 863,000 96.2% 689,000 - 1,036,000 Unspecified signa 34,000 3.8% <500 - 68,000 80 mg strength 313,000 24.2% 208,000 - 417,000 once a day (QD) 313,000 100.0% 208,000 - 417,000 Unspecified drug strength 85,000 6.6% 30,000 - 139,000 M1A Chronic gout 56,000 4.0% 11,000 - 100,000 80 mg strength 56,000 100.0% 11,000 - 100,000 once a day (QD) 56,000 100.0% 11,000 - 100,000 E79 Disorders of purine and pyrimidine metabolism 50,000 3.6% 8,000 - 92,000 40 mg strength 30,000 60.4% <500 - 63,000 once a day (QD) 30,000 100.0% <500 - 63,000 Unspecified drug strength 20,000 39.6% <500 - 46,000 Data Source: Syneos Health Research & Insights Treatment Answers™. 2017. Extracted October 2018. File: Syneos febuxostat & allopurinol by dx, strength & signa 10-30-2018 Note: Syneos Health Research & Insights Treatment Answers™ recommends caution interpreting projected annual uses below 100,000, as the sample size is very small with correspondingly large confidence intervals.

136 of 169 Figure 1. Multivariable Adjusted Hazard Ratios for Any Major CV Event based on all predictors evaluated in the Cox Model in the styudt by Foody et al. (extracted from publication)

137 of 169 Department of Health and Human Services Public Health Service Food and Drug Administration Center for Drug Evaluation and Research Office of Surveillance and Epidemiology Review (OSE) Office of Pharmacovigilance and Epidemiology (OPE)

Epidemiology: Sentinel Memo

Date: 12/10/2018

Reviewer(s): Marie Bradley PhD, MSc.PH, MPharm Division of Epidemiology II Team Leader Efe Eworuke, PhD Division of Epidemiology II Division Director Lockwood Taylor, PhD, MPH Division of Epidemiology II

Drug Name(s): Uloric (febuxostat) Subject Summary of analyses describing gout patients and ULT use in the Sentinel distributed database

Application Type/Number: NDA 021856

Applicant/sponsor: Takeda USA

OSE RCM #: 2018-1542

TSI #: 1849

138 of 169 TABLE OF CONTENTS

EXECUTIVE SUMMARY ...... 2 1 INTRODUCTION ...... 3 2 METHODS USED IN SENTINEL ANALYSES ...... 3 2.1 Data Source: ...... 3 2.2 Characteristics of the gout population ...... 3 2.3 Characteristics of, duration of use and switching patterns among users of febuxostat and allopurinol ...... 4 2.4 Duration of use prior to switching ...... 5 3 SENTINEL ANALYSES RESULTS ...... 6 3.1 Characteristics of the gout population ...... 6 3.2 Characteristics of ULT users, between January 2009- December 2016 ...... 8 3.3 Duration of ULT use ...... 12 3.4 Switching between ULTs ...... 13 3.5 Cumulative duration of use prior to switching ...... 14 4 DISCUSSION ...... 16 5 CONCLUSIONS ...... 18 6 REFERENCES ...... 18 7 Appendix 1 ...... 19

139 of 169 EXECUTIVE SUMMARY

Febuxostat (Uloric®), a selective inhibitor of xanthine oxidase, lowers serum uric acid levels by inhibiting the conversion of xanthine to uric acid. It was approved by the FDA in February 2009 for the management of chronic hyperuricemia in patients with gout. Preliminary results from a post-approval safety trial (Cardiovascular Safety of Febuxostat and Allopurinol in Patients with Gout and Cardiovascular Morbidity (CARES)) showed an increased risk of cardiovascular-related death and all- cause death in febuxostat users. As a result, FDA issued a drug safety communication in November 2017. An advisory committee (AC) meeting is scheduled for January 11, 2019 to discuss potential regulatory action to address the safety of febuxostat. For context, the Division of Pulmonary, Allergy, and Rheumatology Products (DPARP) requested the Division of Epidemiology (DEPI) to investigate the characteristics of the gout population and use of febuxostat and allopurinol in real-world settings using the Sentinel Distributed Database (SDD), since the CARES trial was enriched for patients with CVD.

Between 2009 and 2016, over 5 million gout patients were identified in the SDD, most of whom were male and aged over 65 years. Only a very small proportion of gout patients had a history of CVD at baseline (183 days before or on day of diagnosis) ; for example, only 1.5% had a previous MI and 3% had a prior stroke. Almost half of the gout patients identified in SDD experienced a gout flare defined as two separate dispensings of colchicine, corticosteroids, or prescription non-steroidal anti-inflammatory drug (NSAIDs) between their diagnosis date and end of enrollment. Approximately 30% had gouty arthritis during this time period. Comparing febuxostat and allopurinol initiators in the SDD to the initiators in the CARES trial revealed important differences. Overall ULT users in the CARES trial were younger than in real-world settings with only 48.9% febuxostat and 51.3% allopurinol users aged 65 years or more compared to 66% febuxostat and 64.2% allopurinol in SDD. The CARES trial also had a much higher proportion of male ULT users: 84.1% and 83.8 % of febuxostat and allopurinol users, respectively, compared to 63.6% and 65.2%, respectively, in the SDD. Patients in the CARES trial had more CVD and CKD than those in SDD. For example; 38.6% and 39.8% of febuxostat and allopurinol users, respectively, in CARES compared to 1.5% of allopurinol and febuxostat users, in SDD, had a history of MI. Similarly, 14.8% and 13.3% of users respectively in CARES compared to 2.7% and 2.9% in real world-settings had a prior stroke. In the SDD, few ULT initiators continued use long term (beyond 5+ years) and most remained on the drugs for 1-3 months only and smaller proportion for 1-3 years. Allopurinol users tended to have better adherence, suggested by their slightly longer median duration of use (334 days), compared to febuxostat (210 days). On the contrary, the CARES population appeared to be more adherent with a median duration of exposure to febuxostat of 728 days, and to allopurinol of 719 days. Regarding utilization patterns in the real-world settings, allopurinol (100mg strength), was the most frequently used ULT. The proportion of new users that switched between ULTs during follow-up was low (generally < 10%). The largest proportion (13.4%) of new user switches occurred from allopurinol 100mg to 300mg. Most patients switched after using the initial ULT for 1-3 months. Analyses describing characteristics of ULT initiators in real-world settings revealed important differences when compared to ULT users in the CARES trial. ULT initiators appeared to be older and were less likely to have relevant recent CVD or CKD diagnoses than those in the CARES trial. Adherence was poorer in the real-world settings and switching between ULTs was low. These differences in patient characteristics between real-world users of ULTs and the CARES participants needs to be considered in interpreting the results of the CARES study.

140 of 169 1 INTRODUCTION

Febuxostat (Uloric®), a selective inhibitor of xanthine oxidase lowers serum uric acid levels by inhibiting the conversion of xanthine to uric acid. It was approved by the FDA in February 2009 for the management of chronic hyperuricemia in patients with gout, following three review cycles that included an arthritis advisory committee (AC) meeting, which focused on the drug’s cardiovascular (CV) safety. As a condition of approval, the FDA required a post-market randomized controlled trial (RCT) to determine whether febuxostat use was associated with an increased risk of serious CV outcomes compared to allopurinol. The preliminary results from this study, known as The Cardiovascular Safety of Febuxostat and Allopurinol in Patients with Gout and Cardiovascular Morbidity (CARES) study, released in 2017, showed an increased risk of cardiovascular-related death and all-cause death in febuxostat users. As a result, FDA issued a drug safety communication for febuxostat in November 2017 (https://www.fda.gov/Drugs/DrugSafety/ucm584702.htm). In June 2018, Public Citizen submitted a Citizen’s Petition requesting the immediate removal of febuxostat from the US market based on the results from the CARES study. An advisory committee (AC) meeting is scheduled for January 11, 2019 to discuss potential regulatory action to address the safety of febuxostat. The Division of Pulmonary, Allergy, and Rheumatology Products (DPARP) requested the Division of Epidemiology (DEPI) to investigate characteristics of the gout population and use of febuxostat and allopurinol in real-world settings using the Sentinel Distributed Database (SDD).

2 METHODS USED IN SENTINEL ANALYSES

2.1 DATA SOURCE: The Sentinel System is a distributed network of databases, primarily insurance claims databases. Each partner had medical and pharmacy data, including inpatient and outpatient diagnoses and procedures, and retail and mail order prescription records. Most patients are commercially insured. Within Sentinel’s infrastructure each data partner (DP) maintains physical and operational control over electronic data in their existing environment, with a common data model consisting of standardized administrative and clinical information. Customized pre-tested modular programs compatible with the common data model (referred to as Sentinel’s Active Risk Identification Analysis (ARIA) system) are run on the distributed database. Data from January 1, 2009 to September 30, 2016 from 17 DPs (with over 122 million enrollees) contributing to the Sentinel Distributed Database (SDD) were included in this analysis.

2.2 CHARACTERISTICS OF THE GOUT POPULATION

2.2.1.1 Cohort entry criteria Patients with a diagnosis of gout (ICD-9-CM: 274.xx and ICD-10-CM: M10.xxx, M1A.xxxx in any care setting) between January 1, 2009 and December 31, 2016 were identified. Individuals were required to be continuously enrolled in plans with medical and drug coverage for at least 183 days prior to their gout diagnosis, during which gaps in coverage of up to 45 days were allowed. Given the possibility of misidentifying gout patients by using only one diagnostic code, a sensitivity analysis was performed to include patients with at least two gout diagnoses within 183 days prior to or on the date of the index diagnosis to increase certainty that gout patients identified were true cases.

141 of 169 2.2.1.2 Covariates The following covariates were assessed in the 183 days prior to or on the day of the index gout diagnosis: age (21-44, 45-64, 65+ categories), year, sex, comorbidity score (Combined Comorbidity Index), health service and drug utilization, chronic kidney disease (CKD), and major cardiovascular disease (CVD) as defined in the CARES trial including: myocardial infarction, unstable angina, stroke (hemorrhagic and ischemic), transient ischemic attack, peripheral vascular disease, and diabetes mellitus with evidence of macrovascular or microvascular disease. CKD and CVD were identified by presence of a diagnosis or procedure code in the inpatient care setting, in any diagnosis position. Covariates assessed from the day following the index gout diagnosis to end of enrollment included : ULT use (febuxostat, allopurinol, pegloticase, or probenecid), use of other agents used in the management of gout (colchicine, corticosteroids, and prescription non-steroidal anti-inflammatory drug (NSAIDs), and gout severity measures (presence of tophi, gouty arthritis, kidney stones, or acute gout flares identified separately and in various combinations (e.g., tophi, gouty arthritis, tophi and gouty arthritis, etc.) in any care setting using diagnosis and procedure codes. Gout flares were identified as two separate dispensings of non-ULT gout agents (colchicine, corticosteroids, or prescription non-steroidal anti-inflammatory drug (NSAIDs), on different days, between the day following index gout diagnosis and end of enrollment window on different days.

2.3 CHARACTERISTICS OF, DURATION OF USE AND SWITCHING PATTERNS AMONG USERS OF FEBUXOSTAT AND ALLOPURINOL

2.3.1.1 Cohort entry criteria Among patients with a gout diagnosis, we identified new users of ULTs (febuxostat, allopurinol, pegloticase and probenecid) using National Drug Codes (NDCs). The analysis focused on febuxostat 40mg, febuxostat 80mg, allopurinol 100mg and allopurinol 300mg. In addition, any febuxostat and any allopurinol use was also evaluated separately. Counts of pegloticase and probenecid users were obtained, but no additional information on their characteristics was provided since we expected relatively low use of these products. For patients on each drug and strength, we described the selected baseline characteristics. Gout severity measures and gout flares were assessed in the window from the beginning of enrollment until the index ULT dispensing. Allopurinol and febuxostat users were excluded if they were exposed to any ULT (febuxostat, allopurinol, pegloticase or probenecid) in the 183 days prior to the index date. Pegloticase or probenecid users were excluded if they were exposed to either pegloticase or probenecid only in the 183 days prior to the index date.

2.3.2 Duration of febuxostat or allopurinol use The total cumulative exposure duration was estimated for all new users of febuxostat and allopurinol between January 1, 2009 to December 31, 2016. Febuxostat exposure groups were defined by dispensings of 40 mg strength, 80 mg strength, and any dosage (40 or 80 mg strength). Allopurinol exposure groups were defined by dispensings of 100 mg strength, 300 mg strength, and any dosage (100, 200, or 300 mg strengths and combination products). Exposure episodes were considered continuous if the gap in days’ supply was less than 30 days. Exposure episodes were truncated at the first occurrence of any of the following: 1) the end of the last exposure episode, 2) death, 3) disenrollment, or 4) end of Data Partner data availability. The total length of episodes for each exposure category was reported in categories of <1

142 of 169 month, 1 - <3 months, 3 - <6 months, 6 -<12 months, 1 - <3 years, 3 - <5 years and 5+ years and the Q1, Q3 and median.

2.3.3 Switching between ULTs The switching scenarios listed in Table 1 were evaluated among new allopurinol and febuxostat users. A switch was defined as a new dispensing for the drug of interest following the index dispensing of the prior drug. For example, a switch from febuxostat 40 mg to 80 mg would be captured if new users of febuxostat (40 mg) had a new dispensing of febuxostat (80 mg) during the febuxostat 40mg exposure episode. Exposure episodes were defined using days’ supply from the initial prescription and were considered continuous if gaps in days’ supply were less than 30 days. We also examined the same covariates listed in section 2.1.1.2 for each exposure cohort.

Table 1: Switching scenarios evaluated Initial drug Switch to option 1 Switch to option 2 Switch to option 3 Febuxostat 40 mg Febuxostat 80mg Allopurinol 100mg Allopurinol 300mg Febuxostat 80mg Febuxostat 40 mg Allopurinol 100mg Allopurinol 300mg Febuxostat any Allopurinol any strength strength Allopurinol 100mg Allopurinol 300mg Febuxostat 40 mg Febuxostat 80mg Allopurinol 300mg Allopurinol 100mg Febuxostat 40 mg Febuxostat 80mg Allopurinol any Febuxostat any strength strength

2.4 DURATION OF USE PRIOR TO SWITCHING The cumulative duration of use prior to switching was assessed for the different ULT exposures and drug strengths. For this analysis, we included patients who had a dispensing for the switch of interest after the index dispensing date. We included exposure days past the query end date (December 31, 2016) in the total cumulative exposure duration analysis when there was available data past the query end date (Figure 1). This approach was undertaken to account for scenarios where the switch occurred after the query end date (Figure 1). The duration of use prior to switching was also reported in categories of <1 month, 1 - <3 months, 3 - <6 months, 6 -<12 months, 1 - <3 years, 3 - <5 years and 5+ years and the Q1, Q3 and median.

143 of 169

Figure 1. Cumulative exposure duration of Febuxostat 40 mg with a switch to Allopurinol 100 mg.

1Cumulative exposure duration of Febuxostat 40 mg is truncated at the switch of interest, Allopurinol 100 mg (2/8/2017). By not censoring at the query end date, we capture the total cumulative exposure duration until the switch of interest.

3 SENTINEL ANALYSES RESULTS

3.1 CHARACTERISTICS OF THE GOUT POPULATION There were 5,031,941 patients out of a total eligible 122,135,797 identified in the SDD between January 2009 and December 2016 that had at least one gout diagnosis code. (Table 1). The majority (62.6%) of patients with a gout diagnosis were male and aged 65+ (mean age 67.8 years). Almost half of gout patients (46.4 %) used allopurinol, while only 4.3% used febuxostat. In relation to gout severity markers, just over 50% of gout patients experienced a gout flare (defined as two separate dispensings of colchicine, corticosteroids, or prescription non-steroidal anti-inflammatory drug (NSAIDs) between their diagnosis date and end of enrollment). Around 10% of gout patients had kidney stones, almost 30% had gouty arthritis and 8.3% had tophi between diagnosis date and end of enrollment. Small proportions of the gout population were found to have existing major CVD at baseline (183 days before their first gout diagnosis date); 1.5% had a previous MI, 3% had stroke, 7.3% had diabetes with evidence of micro or macular disease, and 9.3% had CKD. The mean number of unique drug classes was 7.7 (SD, 4.7) and hospitalization and emergency room visits were both rare (Table 2). Although restricting the gout cohort to patients with two or more diagnoses substantially reduced the size, the proportion of males (65.1%) was similar to the previous cohort, with slightly higher mean age (68.9 years). A higher proportion of gout patients in this restricted cohort had ever used allopurinol (60.7%) and febuxostat (6.4%) compared to the original one. For gout severity markers, a similar proportion of gout patients had gout flares (55.6%), kidney stones (10.2%) and tophi (9.7%) in this restricted cohort compared with the original one. Slightly higher proportions of baseline CVD were seen in patients in the

144 of 169 restricted cohort compared to the original cohort; MI 2%, stroke 4% and diabetes evidence of micro or macular disease (9.7%) and 13% CKD. Healthcare utilization was similar between the two cohorts (Table 2).

Table 2: Characteristics of patients with at least one gout diagnosis in the SDD, 2009-2016

Patients with at least one Patients with two or more gout diagnosis gout diagnoses Characteristic N/Mean %/Std Dev2 N/Mean %/Std Dev2 Number of unique patients 5,031,941 2,921,948 Demographics Age (Mean &SD) 67.8 12.2 68.9 12 Age: 21-44 393,716 7.8% 187,409 6.4% Age: 45-64 1,329,986 26.4% 719,919 24.6% Age: 65+ 3,308,239 65.7% 2,014,620 68.9% Gender (Female) 1,881,665 37.4% 1,020,194 34.9% Gender (Male) 3,150,130 62.6% 1,901,674 65.1% Gender (Other) 146 0.0% 80 0.0% Year (2009) 392,773 7.8% 182,919 6.3% Year (2010) 696,542 13.8% 372,129 12.7% Year (2011) 771,471 15.3% 423,296 14.5% Year (2012) 647,867 12.9% 384,385 13.2% Year (2013) 665,844 13.2% 401,162 13.7% Year (2014) 646,719 12.9% 395,014 13.5% Year (2015) 616,698 12.3% 384,306 13.2% Year (2016) 594,027 11.8% 378,737 13.0% Ever use of ULT following gout diagnosis Febuxostat 217,114 4.3% 188,443 6.4% Allopurinol 2,336,478 46.4% 1,772,249 60.7% Probenecid 85,732 1.7% 63,697 2.2% Pegloticase 144 0.0% 143 0.0% Ever use of other gout agents following gout diagnosis Colchicine 1,214,092 24.1% 845,875 28.9% Corticosteroids 2,250,188 44.7% 1,316,508 45.1% Prescription NSAIDs 2,218,766 44.1% 1,200,313 41.1% Gout severity measures following gout diagnosis, through end of enrollment Tophi 416,758 8.3% 282,174 9.7% Gouty arthritis 1,476,568 29.3% 1,075,528 36.8% Kidney stones 513,671 10.2% 296,960 10.2% Gout flares 2,731,306 54.3% 1,624,796 55.6% Tophi and gouty arthritis 215,694 4.3% 171,370 5.9% Tophi and kidney stones 58,145 1.2% 38,375 1.3% Gouty arthritis and kidney stones 172,224 3.4% 125,876 4.3%

145 of 169 Tophi and gouty arthritis and kidney stones 30,317 0.6% 23,717 0.8% History of major CVD in the 183 days before or on day of diagnosis Myocardial infarction 73,669 1.5% 57,296 2.0% Unstable angina 73,556 1.5% 55,524 1.9% Stroke 151,156 3.0% 116,577 4.0% Transient ischemic attack 39,573 0.8% 30,579 1.0% Peripheral vascular disease 185,138 3.7% 146,007 5.0% Diabetic macro- or microvascular disease 369,195 7.3% 284,263 9.7% History of comorbidities in the 183 days before or on day of diagnosis Prior combined comorbidity raw score 1.7 2.7 2.9 2.9 Chronic kidney disease 468,391 9.3% 13.0% 13.0% Health Service Utilization Intensity in the 183 days before or on day of diagnosis Mean number of ambulatory encounters (AV) 10.1 9.3 11.3 9.5 Mean number of emergency room encounters (ED) 0.4 1 0.4 1.1 Mean number of inpatient hospital encounters (IP) 0.3 0.7 0.4 0.9 Mean number of non-acute institutional encounters 0.1 0.5 0.1 0.6 (IS) Mean number of other ambulatory encounters (OA) 3.9 7.7 4.8 8.5 Mean number of unique drug classes 7.7 4.7 8.5 4.7 Mean number of generics 8.2 5.2 9.1 5.3 Mean number of filled prescriptions 22.1 19.5 23.9 19.7

3.2 CHARACTERISTICS OF ULT USERS, BETWEEN JANUARY 2009- DECEMBER 2016 Among patients with a gout diagnosis, we identified 66,682 new febuxostat 40mg, 14,657 new febuxostat 80mg, 681,171 new allopurinol 100mg and 406,323 new allopurinol 300mg users in the SDD between January 2009 and December 2016. The mean age of new users was similar across ULTs ranging from 65- 69 years. A higher proportion of allopurinol 300mg (73.1%) and febuxostat 80mg (72.8%) users were male compared to febuxostat 40mg (61.8%) and allopurinol 100mg (61.7%) (Table 3). Gout severity Some measures of gout severity were seen more frequently in febuxostat users compared to allopurinol users: gout flares febuxostat 40mg (78.8%) and 80 mg (76.9%) compared to allopurinol 100mg (70.3%) and 300mg (67.5%); tophi febuxostat 40mg , (14.3%) and 80mg (17.6%), compared to allopurinol 100mg (9.2%) and 300mg (8.1%); and gouty arthritis febuxostat 40mg (57.6%) and febuxostat 80mg (62.3%) compared to allopurinol 100mg (47.2%) and allopurinol 300mg, (45.4%) [Table 3].

CVD and CKD Allopurinol 300mg and febuxostat 80 mg users were slightly less likely to have diabetes and PVD at baseline compared to allopurinol 100mg users and febuxostat 40mg users. Otherwise, CVD at baseline was mostly similar between febuxostat and allopurinol users (Table 4). Allopurinol users with CKD tended to use the 100mg strength (12.8%) instead of 300mg (6.8%) and a larger proportion of users with CKD used Febuxostat 40mg instead of 80mg.

146 of 169

Healthcare Utilization Healthcare utilization was also largely similar across users (Table 3 and 4), although the mean number of ambulatory encounters and number of drug classes dispensed was slightly higher in the febuxostat groups compared to allopurinol.

Table 3: Characteristics of new initiators of febuxostat and allopurinol by strength in the SDD, 2009-2016

Patients initiating Patients initiating Patients initiating Patients initiating febuxostat febuxostat (80mg) allopurinol (100mg) allopurinol (40mg) (300mg) Characteristic N % N % N % N % Number of unique 66,682 14,657 681,171 406,323 patients Demographics Mean Age (Std Dev) 69 11.8 65 11.8 68.7 12.3 64.9 11.6 Age: 21-44 5,041 7.6 1,595 10.9 51,313 7.5 41,978 10.3 Age: 45-64 16,447 24.7 4,679 31.9 169,497 24.9 129,899 32.0 Age: 65+ 45,194 67.8 8,383 57.2 460,361 67.6 234,446 57.7 Gender (Female) 25,485 38.2 3,990 27.2 260,528 38.2 109,323 26.9 Gender (Male) 41,195 61.8 10,667 72.8 420,621 61.7 296,982 73.1 Gender (Other) 2 0.0 0 0.0 22 0.0 18 0.0 Year (2009) 1,842 2.8 358 2.4 29,106 4.3 24,347 6.0 Year (2010) 5,512 8.3 1,189 8.1 58,578 8.6 41,426 10.2 Year (2011) 9,749 14.6 1,973 13.5 95,903 14.1 62,075 15.3 Year (2012) 9,459 14.2 2,020 13.8 94,912 13.9 58,247 14.3 Year (2013) 10,882 16.3 2,364 16.1 96,774 14.2 55,953 13.8 Year (2014) 9,742 14.6 2,180 14.9 99,806 14.7 56,904 14.0 Year (2015) 9,891 14.8 2,248 15.3 102,005 15.0 54,720 13.5 Year (2016) 9,605 14.4 2,325 15.9 104,087 15.3 52,651 13.0 Baseline cardiovascular history in the 183 days before or on the day of index dispensing Myocardial infarction 1,076 1.6 155 1.1 11,940 1.8 4,462 1.1 Unstable angina 1,084 1.6 203 1.4 11,527 1.7 5,005 1.2 Stroke 1,944 2.9 286 2.0 22,928 3.4 8,838 2.2 Transient ischemic 491 0.7 88 0.6 5,815 0.9 2,402 0.6 attack Peripheral vascular 2,823 4.2 524 3.6 30,571 4.5 11,070 2.7 disease Diabetic macro- or 5,944 8.9 1,089 7.4 60,117 8.8 22,821 5.6 microvascular disease

147 of 169 Baseline gout severity measures at any point before or on the day of index dispensing Tophi 9,555 14.3 2,585 17.6 62,916 9.2 32,901 8.1 Gouty arthritis 38,442 57.6 9,127 62.3 321,743 47.2 184,416 45.4 Kidney stones 7,130 10.7 1,609 11.0 60,287 8.9 36,084 8.9 Gout flares 51,270 76.9 11,554 78.8 478,559 70.3 274,178 67.5 Tophi and gouty 7,361 11.0 2,172 14.8 41,300 6.1 21,075 5.2 arthritis Tophi and kidney 1,212 1.8 309 2.1 6,703 1.0 3,435 0.8 stones Gouty arthritis and 4,278 6.4 1,046 7.1 29,359 4.3 16,389 4.0 kidney stones Tophi and gouty 962 1.4 264 1.8 4,416 0.6 2,179 0.5 arthritis and kidney stones Baseline history in the 183 days before or on the day of index dispensing Prior combined 2.4 2.8 2 2.7 2.1 2.9 1.2 2.4 comorbidity raw score Chronic kidney disease 9,504 14.3 1,622 11.1 87,240 12.8 27,428 6.8 Health Service Utilization Intensity in the 183 days before or on the day of index dispensing Mean number of 12.7 10 11.7 10 11 9.2 9.4 8.4 ambulatory encounters (AV) Mean number of 0.4 0.9 0.4 1.1 0.4 1 0.4 1 emergency room encounters (ED) Mean number of 0.3 0.8 0.3 0.7 0.3 0.8 0.2 0.7 inpatient hospital encounters (IP) Mean number of non- 0.1 0.5 0.1 0.5 0.1 0.6 0.1 0.5 acute institutional encounters (IS) Mean number of other 4.7 8.4 3.9 7.3 5.1 9.4 3.3 6.7 ambulatory encounters (OA) Mean number of 10.3 4.8 9.6 4.7 9.6 4.6 8.5 4.4 unique drug classes Mean number of 11.1 5.4 10.3 5.3 10.2 5.2 9.1 4.9 generics Mean number of filled 27.3 20.1 24.5 18.6 24.3 18.8 20.5 16.6 prescriptions

148 of 169

Table 4: Characteristics of new initiators of febuxostat and allopurinol combined in the SDD, 2009- 2016

Febuxostat initiators Allopurinol initiators Characteristic1 N/Mean %/Std Dev2 N/Mean %/Std Dev2 Number of unique patients 80,083 1,049,462 Demographics Mean Age, Std Dev 68.3 11.9 67.3 12.1 Age: 21-44 6,503 8.1 88,682 8.5 Age: 45-64 20,722 25.9 286,874 27.3 Age: 65+ 52,858 66.0 673,906 64.2 Gender (Female) 29,120 36.4 359,524 34.3 Gender (Male) 50,961 63.6 689,900 65.7 Gender (Other) 2 0.0 38 0.0 Year (2009) 2,198 2.7 52,761 5.0 Year (2010) 6,674 8.3 98,425 9.4 Year (2011) 11,655 14.6 154,836 14.8 Year (2012) 11,346 14.2 148,535 14.2 Year (2013) 13,060 16.3 146,940 14.0 Year (2014) 11,688 14.6 149,906 14.3 Year (2015) 11,839 14.8 149,287 14.2 Year (2016) 11,623 14.5 148,772 14.2 Baseline cardiovascular history in the 183 days before or on the day of index dispensing Myocardial infarction 1,204 1.5 15,897 1.5 Unstable angina 1,267 1.6 16,017 1.5 Stroke 2,197 2.7 30,870 2.9 Transient ischemic attack 570 0.7 7,984 0.8 Peripheral vascular disease 3,288 4.1 40,445 3.9 Diabetic macro- or 6,926 8.6 80,595 7.7 microvascular disease Baseline gout severity measures at any point before or on the day of index dispensing Tophi 11,816 14.8 90,446 8.6 Gouty arthritis 46,593 58.2 481,661 45.9 Kidney stones 8,527 10.6 92,022 8.8 Gout flares 61,655 77.0 719,277 68.5 Tophi and gouty arthritis 9,226 11.5 57,979 5.5 Tophi and kidney stones 1,462 1.8 9,414 0.9 Gouty arthritis and kidney 5,149 6.4 42,766 4.1 stones Tophi and gouty arthritis and 1,169 1.5 5,982 0.6 kidney stones Baseline history in the 183 days before or on the day of index dispensing

149 of 169 Prior combined comorbidity 2.4 2.8 1.8 2.8 raw score Chronic kidney disease 10,958 13.7 111,424 10.6 Health Service Utilization Intensity in the 183 days before or on the day of index dispensing Mean number of ambulatory 12.5 10 10.5 9 encounters (AV), Std Dev Mean number of emergency 0.4 1 0.4 1 room encounters (ED), Std Dev Mean number of inpatient 0.3 0.8 0.3 0.8 hospital encounters (IP), Std Dev Mean number of non-acute 0.1 0.5 0.1 0.6 institutional encounters (IS), Std Dev Mean number of other 4.6 8.2 4.4 8.5 ambulatory encounters (OA), Std Dev Mean number of unique drug 10.2 4.7 9.2 4.5 classes, Std Dev Mean number of generics, Std 10.9 5.4 9.8 5.1 Dev Mean number of filled 26.8 19.9 23 18.2 prescriptions, Std Dev

3.3 DURATION OF ULT USE Table 5 shows cumulative exposure duration for new febuxostat and allopurinol users in SDD between 2009 and 2016. Overall, almost 30% of febuxostat and 22% allopurinol initiators continue using these agents for approximately 1-3 months, 26-30% use the drugs for 1-3 years, and only very small proportions of ULT users continuing long term use beyond 5 years. Across duration of use categories (<1 month, 1-3 months, 3-6 months, 6 months- 1 year 1-3 years, 3- <5 years and 5 years+) proportions of male and female users tend to be fairly consistent with approximately two thirds being male and one third female in most categories. (Appendix 1, Table 1). Allopurinol users had longer median durations of use at 216 and 270 days for allopurinol 100 and 300mg users respectively compared to 165 and 180 days for febuxostat 40mg and 80mg users (Table 5). Females had longer median duration of use than males for all agents and strengths examined except allopurinol 300mg (Appendix 1, Table 1). Those aged 65+ tended to use ULTs for longer durations than younger users (Appendix 1 Table 2)

150 of 169 Table 5: Distribution of cumulative exposure duration and median duration of use for all ULTs examined in the SDD, 2009-2016

<1 1 - <3 3 - <6 6 - <12 1 - <3 3 - <5 5+ Q1 Medi Q3 month month month month years years years (days) an (days) s s s (days)

Exposures % % % % % % %

Febuxostat 40 mg 3.5 34.4 15 15.2 23.2 6.8 1.9 60 165 496 (n=66,682)

Febuxostat 80 mg 2.9 32.8 16.2 17.3 23.8 5.7 1.3 60 180 460 (n=14,657)

Febuxostat -any 3.2 29.6 14.1 15.7 26.7 8.3 2.4 62 210 600 strength (n=80,083)

Allopurinol 100 4.4 27.4 14.2 15.4 26.1 9.2 3.3 75 216 637 mg (n=681,171) Allopurinol 300 2.6 26 13.1 15.2 28 10.8 4.3 90 270 745 mg (n= 406,322) Allopurinol any 2.8 21.9 12.4 15.1 30.4 12.4 5 93 334 839 strength- (n=1,049,461)

3.4 SWITCHING BETWEEN ULTS Table 6 shows the switching patterns of febuxostat and allopurinol users. Overall, the proportion of new users that switched between ULTs during follow-up was low (generally < 10%). The largest proportion of new users’ switches occurred from allopurinol 100mg to 300mg (13.4% of new users switched) febuxostat 40mg to febuxostat 80mg (9.1% new users switched). The next most common switches were of allopurinol 300mg to allopurinol 100mg (6.8%) and febuxostat 40mg new users switched to allopurinol 100mg (6.3%). Other switch combinations were not common. Febuxostat 80 mg users most commonly switched to febuxostat 40mg while febuxostat 40mg users were most likely to switch to febuxostat 80mg. Gender No consistent switching patterns were observed with respect to gender across the studied switching scenarios; however, women on any strength of febuxostat were slightly more likely to switch to any strength of allopurinol compared to men (10.1% vs 9.5% for women and men respectively). Similarly, women on any strength of allopurinol were more likely than men to switch to any strength of febuxostat (3% vs 2.5% for women vs men respectively [Table 6].) Age

151 of 169 Those aged 65+ were also slightly more to switch compared to younger users aged 21-44 and 45-64 (10.2% vs 8% and 8.8% respectively) for switching from any strength febuxostat to any strength allopurinol and a similar trend was observed in those switching from any strength of allopurinol to any strength of febuxostat 2.9% vs 1.9% and 2.3% respectively (Table 6).

Table 6. Proportion of new users of febuxostat and allopurinol that switched, by gender and sex

Switching Scenario Overall a Females Males 21-44 45-64 65+ (%)a (%)b (%)b years years years (%)c (%)c (%)c Febuxostat (80mg) users that Switch to 5.1 6.9 4.5 2.6 4.5 5.9 Febuxostat (40mg) Febuxostat (80mg) users 3.3 3.9 3.1 2.0 2.8 3.9 s that Switch to Allopurinol (100mg) Febuxostat (80mg) users 4.2 3.9 4.3 4.3 4.1 4.1 Switch to Allopurinol (300mg) Febuxostat (40mg) users 9.1 7.5 10.1 10.3 10.1 8.6 that Switch to Febuxostat (80mg) Febuxostat (40mg) users that Switch to 6.3 7.3 5.6 3.8 5.0 7.0 Allopurinol (100mg) Febuxostat (40mg) users that Switch to 3.5 2.9 3.9 4.2 3.8 3.3 Allopurinol (300mg) Allopurinol (100mg) users 13.4 10.6 15.2 16.2 15.2 12.3 that Switch to Allopurinol (300mg) Allopurinol (100mg) users that Switch t 2.1 2.4 1.9 1.3 1.6 2.4 o Febuxostat (40mg) Allopurinol (100mg) users 0.4 0.4 0.4 0.3 0.4 0.4 that Switch to Febuxostat (80mg) Allopurinol (300mg) users that Switch t 6.8 7.9 6.4 6.2 5.8 7.4 o Allopurinol (100mg) Allopurinol (300mg) users 1.7 2.2 1.5 1.3 1.5 1.9 Switch to Febuxostat (40mg) Allopurinol (300mg users 0.7 0.7 0.7 0.7 0.7 0.7 Switch to Febuxostat (80mg) Febuxostat users of Any Dose that Swit 9.7 10.1 9.5 8.0 8.8 10.2 ch to Any Allopurinol Dose Allopurinol users of Any Dose that Swit 2.7 3.0 2.5 1.9 2.3 2.9 ch to Any Febuxostat Dose a Overall proportion of new users that switched. For example; for the switching scenario “Febuxostat 80mg users that switched to febuxostat 40mg” there were 14,657 new febuxostat 80mg users and 754 of these switched to febuxostat 40mg.754/14657 = 5.1% b Proportion of female new users that switched. For example; for the switching scenario “Febuxostat 80mg users that switched to febuxostat 40mg” there were 3990 females and 10,667 males who initiated febuxostat 80mg and 277 females and 477 males that switched to febuxostat 40mg. 277/3990 = 6.9% c Proportion of new users aged 21-44 who switched. For example; for the switching scenario “Febuxostat 80mg users that switched to febuxostat 40mg” there were 1595 patients aged 21-44 years who initiated febuxostat 80mg and 41 of these new users switched to febuxostat 40mg. 41/1595 = 2.6%

3.5 CUMULATIVE DURATION OF USE PRIOR TO SWITCHING The cumulative durations of ULT use before switching to another strength or drug are shown in Table 7.

152 of 169

Febuxostat users (any strength) who switched to allopurinol (any strength) Almost 40% of users remained on their initial febuxostat prescription for 1-3 months before switching. The proportions of patients that switched agents after 3-6 months (15.1%), 6 months-1 year (14.2%) and 1-3 years (17.7%) of use were similar. Smaller proportions of patients (11.7%) switched after less than one month and the least number switched after long term use 3-<5 years (3.5%).

Allopurinol users (any strength) that switched to Febuxostat (any strength) Most allopurinol users switched after 1-3 months (31.4%) with the smallest proportion of switches occurring within 1 month of initiating the allopurinol (9.3%) and after long term use (3-<5 years [4.7%]). Febuxostat 80 mg, to Febuxostat 40 mg and Allopurinol 300 mg, to Allopurinol 100 mg A larger proportion of switches occurred in the 1-3 year period after initial use among those switching from febuxostat 80 mg, to febuxostat 40 mg (23.5%) and from allopurinol 300 mg, switch to allopurinol 100 mg (23.3%) compared to other switching scenarios (Table 7). Age Among switchers from febuxostat (any strength) to allopurinol (any strength) and allopurinol (any strength) to febuxostat (any strength), no major differences were observed in the duration of use before switching by age group 21-44 and 45-64 and 65+. (Appendix 1- Table 4)

3.5.1 Median time to switch Among febuxostat users (any strength) who switched to allopurinol (any strength), the median time to switch was 95 days, and for allopurinol users (any strength) switching to febuxostat (any strength), it was 134 days. The longest median time to switch (173 days) occurred in those on febuxostat 80 mg who switched to febuxostat 40 mg (Table 7). Gender There was very little difference in median duration of use before switching between women and men ( 91 vs 96 days, respectively) among febuxostat users (any strength) who switched to allopurinol (any strength). However, among allopurinol users (any strength) who switched to febuxostat (any strength), men tended to have a longer duration of use prior to switching than women (150 vs 116 days respectively) (Appendix 1- Table 3). Age Median duration of use before switching was slightly longer in those aged 65+ (100 days), compared to those aged age 21-44 and 45-64 (78 days and 92 days respectively) among switchers from febuxostat (any strength) to allopurinol. Among those switching from allopurinol (any strength) to febuxostat (any strength), only minor differences in mean duration of use before switching according to age were observed (134, 133 and 136 days respectively). (Appendix – Table 4)

153 of 169 Table 7: Distribution of cumulative exposure duration among patients who switched ULTs in the SDD, 2009-2016

<1 1 - <3 3 - <6 6 months - 1 - <3 3 - <5 5+ Q1 Median Q3 month months months <12 years years years time to months switch Exposures % % % % % % % (days) (days) (days) Febuxostat 40 mg, switch to 9.4 36.2 20 16.1 15.4 2.5 0.3 46 108 267 Febuxostat 80 mg (n= 9,598) Febuxostat 40 mg, switch to 12.6 41.5 15 13.1 14.8 2.6 0.4 30 90 244 Allopurinol 100 mg (n=11,354) Febuxostat 40 mg, switch to 9.7 45 16.1 12.9 13.7 2.3 0.3 30 90 219 Allopurinol 300 mg (n=7,464) Febuxostat 80 mg, switch to 6.6 29.1 18 17.5 23.5 4.5 0.9 60 173 433 Febuxostat 40 mg (n=1,994) Febuxostat 80 mg, switch to 8.9 42 16.2 14.8 15.8 1.9 0.4 30 90 264 Allopurinol 100 mg (n=1,836) Febuxostat 80 mg, switch to 8.4 43.3 15.8 14 16 2.3 0.3 30 90 270 Allopurinol 300 mg (n=2,066) Febuxostat (any strength), 11.7 37.3 15.1 14.2 17.7 3.5 0.5 30 95 308 switch to Allopurinol (any strength) (n=19,363) Allopurinol 100 mg, switch 15.2 32.1 18.2 15.3 15.5 3 0.5 32 100 272 to Allopurinol 300 mg (n=145,823 Allopurinol 100 mg, switch 11.8 37.4 17.4 14.9 15.1 30 92 30 93 270 to Febuxostat 40 mg Allopurinol 100 mg, switch 7 42.6 18.6 15.5 14.1 48 165 30 92 245 to Febuxostat 80 mg Allopurinol 300 mg, switch 12.7 25.9 14.5 15.7 23.3 31 97 48 165 472 to Allopurinol 100 mg Allopurinol 300 mg, switch 9.3 39.1 16.9 14.3 16.4 57 120 31 97 285 to Febuxostat 40 mg Allopurinol 300 mg, switch 6 38.7 18.2 16 17.1 53 134 57 120 304 to Febuxostat 80 mg Allopurinol (any strength) 9.3 31.4 17.1 16.3 20.2 4.7 1 53 134 378 switch to Febuxostat (any strength)

4 DISCUSSION

The analyses described in this report were conducted to evaluate characteristics of the gout population and ULT users, to describe ULT use in the real-world settings, and to compare this population to those included in the CARES trial. Allopurinol was the most commonly used ULT among gout patients and the 100mg strength was most often used. Febuxostat use was comparatively rare with just 4.3% of gout patients initiating it in the study

154 of 169 period. Febuxostat initiators were slightly older than allopurinol initiators (mean age 68 vs. 67 years), and the proportion of initiators aged > 65 years was 66% and 64.2%, respectively. More women initiated febuxostat than allopurinol (36.4%vs 34.3% respectively). In comparison in the CARES trial those assigned to allopurinol were slightly older than those to febuxostat (mean age 65 vs 64 years respectively) and overall ULT users in CARES were younger than in real-world settings with only 48.9% febuxostat and 51.3% allopurinol users aged 65 years or more. The CARES trial also had a much higher proportion of male ULT users than in real-world settings with 84.1% and 83.8 % of those in the febuxostat and allopurinol arms respectively classed as male. Patients in the CARES trial had more CVD and CKD than those in SDD. For example; 38.6% and 39.8% of febuxostat and allopurinol users, respectively, in CARES compared to 1.5% of allopurinol and febuxostat users, in SDD, had a history of MI. Similarly, 14.8% and 13.3% of users respectively in CARES compared to 2.7% and 2.9% in real world-settings had a prior stroke. However, this was expected as the CARES trial was enriched for CVD. Those with CKD, diabetes and PVD in real-world settings tended to use lower strengths of both allopurinol and febuxostat, which at least for CKD, was expected. In real-world settings, febuxostat users tended to have more severe gout than allopurinol users indicated by the frequency of gout flares (77% vs 68.5% respectively) and occurrence of tophi (14.8% vs 8.6% respectively) and gouty arthritis (58.2 % vs 45.9% respectively). The higher cost of febuxostat 1 may limit its use to those with more severe gout in real- world settings. As allopurinol 100mg was the most commonly used ULT in real-world settings, it may appear that gout patients are being under-strengthd given the recommended starting strength for alopurinol is 300mg. However, when switching between ULTs was examined, we found that switching from allopurinol 100mg to 300mg was the most common switching scenario indicating that patients in real-world settings may be titrating slowly to the optimal allopurinol strength.

In real-world settings, few ULT initiators continued use long term (beyond 5+ years) and most remained on the drugs for 1-3 months only and smaller proportion for 1-3 years. Allopurinol users tended to have better adherence, suggested by their longer median duration of use (334 days), compared to febuxostat (210 days). Female gout patients tended to adhere slightly better to all ULTs than male patients. Those aged 65+ also appeared to adhere better compared to younger users. The CARES population appeared to be more adherent than those in real-world settings with a median duration of exposure to febuxostat of 728 days, and to allopurinol of 719 days. Overall in CARES, 56.6% of patients discontinued ULT treatment prematurely and the rates of premature discontinuation were similar in the febuxostat and allopurinol groups (57.3% and 55.9%, respectively). The low adherence in real-world settings is in line with the poor adherence rates for ULTs in gout patients that have been described previously2 The proportion of new users that switched between ULTs during follow-up, in real-world settings, was low. The largest proportion (13.4%) of new user switches occurred from allopurinol 100mg to 300mg which as stated above may suggest slow strength titration in some allopurinol users. The next most common switches were from febuxostat 40mg to febuxostat 80mg, allopurinol 300mg to allopurinol 100mg and febuxostat 40mg to allopurinol 100mg. The proportion of switches from allopurinol any strength to febuxostat any strength was low (2.7%). Most patients switched after using the initial ULT for 1-3 months. New users of febuxostat who did switch to allopurinol tended to do so more quickly than new users of allopurinol who switched to febuxostat. This study was large and included over 5 million gout patients from seventeen diverse DPs contributing to the SDD from across the US. The SDD now contains data on those aged 65 years+, which makes the findings more generalizable to the US population. The study also had some limitations. CVD prevalence in study participants was likely underestimated as we had a limited look back period to determine prior diagnoses (one year prior to cohort entry). We did not have information on formulary restrictions related to ULT use for DPs in our study. However, we included seventeen DPs with diverse formularies which should reflect practice in real-world settings.

155 of 169

5 CONCLUSIONS Analyses describing characteristics of ULT initiators in real-world settings revealed important differences when compared to ULT users in the CARES trial. Although in real-world settings ULT initiators appeared to be older, they were less likely to have relevant recent CVD or CKD diagnoses than those in the CARES trial. Adherence was poorer in the real-world settings and switching between ULTs was low. The differences in patient characteristics between real-world users of ULTs and the CARES participants needs to be considered in interpreting the results of the CARES study.

6 REFERENCES

1. Jutkowitz E, Choi HK, Pizzi LT, Kuntz KM. Cost-Effectiveness of Allopurinol and Febuxostat for the Management of Gout Ann Intern Med. 2014 Nov 4;161(9):617-26.

2. Yin R, Li L, Zhang G, Cui Y, Zhang L, Zhang Q, Fu T, Cao H, Li L, Gu Z.Rate of adherence to urate-lowering therapy among patients with gout: a systematic review and meta-analysis. BMJ Open. 2018 Apr 10;8(4):e017542.

156 of 169

7 APPENDIX 1 Table 1: Distribution of cumulative exposure duration, by sex, in the SDD, 2009-2016 <1 month 1 - <3 months 3 - <6 months 6 months - <1 1 - <3 years 3 - <5 years 5+ years Q1 Media Q3 year (day n (days) s) (days) Exposures N % N % N % N % N % N % N % (n) Febuxostat 2,351 100 22,911 100 10,034 100 10,106 100 15,498 100 4,535 100 1,247 100 60 496 40 mg (66,682) Female 946 40.2 8,225 35.9 3,647 36.3 3,880 38.4 6,277 40.5 2,009 44.3 501 40.2 60 181 545 (25,485) Male 1,405 59.8 14,685 64.1 6,386 63.6 6,226 61.6 9,221 59.5 2,526 55.7 746 59.8 60 153 465 (41,195) Other 2 0 0 1 0 1 0 0 0 0 0 0 0 0 0 30 105 180 Febuxostat 432 100 4,810 100 2,376 100 2,531 100 3,485 100 829 100 194 100 60 180 460 80 mg (14,657) Female 125 28.9 1,267 26.3 586 24.7 713 28.2 976 28 256 30.9 67 34.5 60 184 488 (3,990) Male 307 71.1 3,543 73.7 1,790 75.3 1,818 71.8 2,509 72 573 69.1 127 65.5 60 172 450 (10,667) Other 0 0 0 0 0 0 0 0 0 0 0 0 0 0 - - - Febuxostat 2,574 100 23,698 100 11,286 100 12,566 100 21,383 100 6,685 100 1,891 100 62 210 600 (any strength) (80,083) Female 1,010 39.2 8,256 34.8 3,886 34.4 4,549 36.2 8,006 37.4 2,725 40.8 688 36.4 65 229 640 (29,120)

157 of 169 Male 1,564 60.8 15,442 65.2 7,399 65.6 8,017 63.8 13,376 62.6 3,960 59.2 1,203 63.6 61 203 575 (50,961) Other 0 0 0 0 1 0 0 0 1 0 0 0 0 0 180 345 510 Allopurinol 30,203 100 186,65 100 96,870 100 105,20 100 177,50 100 62,56 100 22,17 100 75 216 637 100 mg 1 6 0 3 8 (681,171) Female 10,536 34.9 67,368 36.1 35,044 36.2 39,304 37.4 72,202 40.7 27,09 43.3 8,984 40.5 90 254 703 (260,528) 0 Male 19,667 65.1 119,28 63.9 61,825 63.8 65,895 62.6 105,29 59.3 35,47 56.7 13,19 59.5 67 201 597 (420,621) 0 1 0 3 Other 0 0 3 0 1 0 7 0 7 0 3 0 1 0 237 357 858 Allopurinol 10,580 100 105,69 100 53,066 100 61,888 100 113,61 100 43,92 100 17,54 100 90 270 745 300 mg 7 9 8 4 (406,322) Female 3,023 28.6 30,505 28.9 13,876 26.1 15,862 25.6 29,759 26.2 11,99 27.3 4,301 24.5 90 261 733 (109,322) 6 Male 7,556 71.4 75,189 71.1 39,186 73.8 46,023 74.4 83,856 73.8 31,93 72.7 13,24 75.5 90 276 749 (296,982) 0 2 Other (18) 1 0 3 0 4 0 3 0 4 0 2 0 1 0 100 273 767 Allopurinol 29,186 100 229,53 100 129,82 100 158,95 100 319,25 100 130,4 100 52,28 100 93 334 839 (any 6 7 5 8 17 2 strength) (1,049,461) Female 10,203 35 79,081 34.5 43,193 33.3 52,690 33.1 110,32 34.6 46,96 36 17,06 32.6 92 345 856 (359,523) 1 7 8 Male 18,982 65 150,44 65.5 86,631 66.7 106,25 66.8 208,92 65.4 83,44 64 35,21 67.4 94 330 830 (689,900) 9 7 4 5 2 Other 1 0 6 0 3 0 8 0 13 0 5 0 2 0 181 397 858 (38)

158 of 169

Table 2: Distribution of cumulative exposure duration by age group, in the SDD, 2009-2016

<1 month 1 - <3 months 3 - <6 months 6 months - <1 1 - <3 years 3 - <5 years 5+ years Q1 Median Q3 year Exposures N % N % N % N % N % N % N % days days days (n) Febuxostat 2,351 100 22,911 100 10,034 100 10,106 100 15,498 100 4,535 100 1,247 100 165 496 3,246 40 mg

(66,682) 21-44 180 7.7 2,334 10.2 775 7.7 681 6.7 812 5.2 186 4.1 73 5.9 92 298 3,099 (5,041) 45-64 583 24.8 6,203 27.1 2,561 25.5 2,398 23.7 3,462 22.3 916 20.2 324 26 135 433 3,246 (16,447) 65+ 1,588 67.5 14,374 62.7 6,698 66.8 7,027 69.5 11,224 72.4 3,433 75.7 850 68.2 182 542 3,184 (45,194) Febuxostat 432 100 4,810 100 2,376 100 2,531 100 3,485 100 829 100 194 100 180 460 3,098 80 mg (14,657) 21-44 59 13.7 610 12.7 277 11.7 266 10.5 313 9 56 6.8 14 7.2 129 354 2,642 (1,595) 45-64 124 28.7 1,686 35.1 757 31.9 785 31 1,028 29.5 230 27.7 69 35.6 151 428 2,955 (4,679) 65+ 249 57.6 2,514 52.3 1,342 56.5 1,480 58.5 2,144 61.5 543 65.5 111 57.2 190 506 3,098 (8,383) Febuxostat 2,574 100 23,698 100 11,286 100 12,566 100 21,383 100 6,685 100 1,891 100 210 600 3,274 (any strength) (80,083) 21-44 226 8.8 2,515 10.6 972 8.6 966 7.7 1,367 6.4 333 5 124 6.6 135 420 3,099 (6,503)

159 of 169 45-64 657 25.5 6,690 28.2 3,003 26.6 3,176 25.3 5,161 24.1 1,483 22.2 552 29.2 183 547 3,246 (20,722) 65+ 1,691 65.7 14,493 61.2 7,311 64.8 8,424 67 14,855 69.5 4,869 72.8 1,215 64.3 238 640 3,274 (52,858) Allopurinol 30,203 100 186,651 100 96,870 100 105,206 100 177,500 100 62,563 100 22,178 100 216 637 3,463 100 mg (681,171) 21-44 3,074 10.2 20,152 10.8 8,709 9 7,616 7.2 8,865 5 2,143 3.4 754 3.4 107 327 3,275 (51,313) 45-64 8,113 26.9 51,612 27.7 26,431 27.3 26,809 25.5 39,781 22.4 12,157 19.4 4,594 20.7 180 523 3,450 (169,497) 65+ 19,016 63 114,887 61.6 61,730 63.7 70,781 67.3 128,854 72.6 48,263 77.1 16,830 75.9 267 716 3,463 (460,361) Allopurinol 10,580 100 105,697 100 53,066 100 61,888 100 113,619 100 43,928 100 17,544 100 270 745 3,441 300 mg (406,322) 21-44 1,267 12 14,475 13.7 6,527 12.3 6,665 10.8 9,351 8.2 2,635 6 1,058 6 164 473 3,329 (41,978) 45-64 3,267 30.9 35,078 33.2 18,400 34.7 20,722 33.5 35,143 30.9 11,997 27.3 5,292 30.2 243 669 3,441 (129,899) 65+ 6,046 57.1 56,144 53.1 28,139 53 34,501 55.7 69,125 60.8 29,296 66.7 11,194 63.8 318 832 3,441 (234,445) Allopurinol 29,186 100 229,536 100 129,827 100 158,955 100 319,258 100 130,417 100 52,282 100 334 839 3,464 (any strength) (1,049,461) 21-44 2,915 10 27,643 12 13,349 10.3 14,078 8.9 21,626 6.8 6,464 5 2,607 5 187 543 3,364 (88,682) 45-64 7,815 26.8 67,670 29.5 38,460 29.6 45,352 28.5 83,457 26.1 30,504 23.4 13,616 26 292 755 3,464 (286,874) 65+ 18,456 63.2 134,223 58.5 78,018 60.1 99,525 62.6 214,175 67.1 93,449 71.7 36,059 69 381 907 3,463 (673,905)

160 of 169

Table 3. Distribution of cumulative exposure duration, among patients with evidence of switch, by sex, in the SDD, 2009-2016 <1 month 1 - <3 3 - <6 6 months - 1 - <3 years 3 - <5 years 5+ years Q1 Median Q3 months months <1 year Exposures (n) N % N % N % N % N % N % N % days days days Febuxostat 40 mg, 902 100 3,475 100 1,920 100 1,548 100 1,479 100 242 100 32 100 46 108 267 switch to Febuxostat 80 mg (9,598) Female (2,980) 275 30.5 1,052 30.3 567 29.5 476 30.7 518 35 81 33.5 11 34.4 49 115 294 Male (6,617) 626 69.4 2,423 69.7 1,353 70.5 1,072 69.3 961 65 161 66.5 21 65.6 45 104 253 Other (1) 1 0.1 0 0 0 0 0 0 0 0 0 0 0 0 27 27 27 Febuxostat 40 mg, 1,433 100 4,712 100 1,703 100 1,492 100 1,682 100 292 100 40 100 30 90 244 switch to Allopurinol 100 mg (11,354) Female (4,626) 661 46.1 1,809 38.4 690 40.5 583 39.1 733 43.6 131 44.9 19 47.5 30 90 267 Male (6,727) 772 53.9 2,902 61.6 1,013 59.5 909 60.9 949 56.4 161 55.1 21 52.5 30 90 232 Other ( 1) 0 0 1 0 0 0 0 0 0 0 0 0 0 0 30 30 30 Febuxostat 40 mg, 723 100 3,359 100 1,201 100 961 100 1,023 100 174 100 23 100 30 90 219 switch to Allopurinol 300 mg (7,464) Female (2,110) 229 31.7 905 26.9 348 29 252 26.2 322 31.5 50 28.7 4 17.4 30 90 237 Male (5,354) 494 68.3 2,454 73.1 853 71 709 73.8 701 68.5 124 71.3 19 82.6 30 90 215 Other (0) 0 0 0 0 0 0 0 0 0 0 0 0 0 0 - - - Febuxostat 80 mg, 132 100 581 100 358 100 348 100 468 100 90 100 17 100 60 173 433 switch to Febuxostat 40 mg (1,994)

161 of 169 Female (676) 39 29.5 198 34.1 94 26.3 125 35.9 176 37.6 37 41.1 7 41.2 61 186 494 Male (1,318) 93 70.5 383 65.9 264 73.7 223 64.1 292 62.4 53 58.9 10 58.8 60 158 390 Other ( 0) 0 0 0 0 0 0 0 0 0 0 0 0 0 0 - - - Febuxostat 80 mg, 164 100 772 100 297 100 271 100 291 100 34 100 7 100 30 90 264 switch to Allopurinol 100 mg (1,836) Female (550) 56 34.1 238 30.8 77 25.9 83 30.6 85 29.2 10 29.4 1 14.3 30 90 259 Male (1,286) 108 65.9 534 69.2 220 74.1 188 69.4 206 70.8 24 70.6 6 85.7 30 91 270 Other ( 0) 0 0 0 0 0 0 0 0 0 0 0 0 0 0 - - - Febuxostat 80 mg, 173 100 894 100 326 100 289 100 331 100 47 100 6 100 30 90 270 switch to Allopurinol 300 mg (2066) Female (444) 53 30.6 183 20.5 50 15.3 65 22.5 77 23.3 15 31.9 1 16.7 30 90 297 Male (1,622) 120 69.4 711 79.5 276 84.7 224 77.5 254 76.7 32 68.1 5 83.3 30 90 253 Other (0) 0 0 0 0 0 0 0 0 0 0 0 0 0 0 - - - Febuxostat (any 2,261 100 7,220 100 2,930 100 2,747 100 3,433 100 671 100 101 100 30 95 308 strength)2, switch to Allopurinol (any strength)3 (19,363) Female (6,745) 928 41 2,423 33.6 968 33 917 33.4 1,242 36.2 237 35.3 30 29.7 30 91 320 Male (12,617) 1,333 59 4,797 66.4 1,962 67 1,830 66.6 2,190 63.8 434 64.7 71 70.3 30 96 304 Other (1) 0 0 0 0 0 0 0 0 1 0 0 0 0 0 510 510 510 Allopurinol 100 mg, 22,217 100 46,871 100 26,567 100 22,379 100 22,660 100 4,364 100 765 100 32 100 272 switch to Allopurinol 300 mg (145,823) Female (42,738) 6,362 28.6 13,010 27.8 7,401 27.9 6,584 29.4 7,530 33.2 1,606 36.8 245 32 36 110 313 Male (103,083) 15,855 71.4 33,861 72.2 19,166 72.1 15,793 70.6 15,130 66.8 2,758 63.2 520 68 31 97 259 Other (2) 0 0 0 0 0 0 2 0 0 0 0 0 0 0 237 259 281

162 of 169 Allopurinol 100 mg, 3,785 100 11,962 100 5,562 100 4,786 100 4,820 100 924 100 183 100 30 93 270 switch to Febuxostat 40 mg (32,022) Female (13,233) 1,710 45.2 4,978 41.6 2,207 39.7 1,863 38.9 1,979 41.1 425 46 71 38.8 30 90 270 Male (18,787) 2,075 54.8 6,984 58.4 3,355 60.3 2,921 61 2,841 58.9 499 54 112 61.2 30 99 269 Other (2) 0 0 0 0 0 0 2 0 0 0 0 0 0 0 237 276 314 Allopurinol 100 mg, 726 100 4,390 100 1,921 100 1,594 100 1,453 100 200 100 27 100 30 92 245 switch to Febuxostat 80 mg (10,311) Female (3,284) 248 34.2 1,417 32.3 563 29.3 495 31.1 481 33.1 71 35.5 9 33.3 30 90 257 Male (7,026) 478 65.8 2,973 67.7 1,358 70.7 1,098 68.9 972 66.9 129 64.5 18 66.7 30 93 241 Other (1) 0 0 0 0 0 0 1 0.1 0 0 0 0 0 0 237 237 237 Allopurinol 300 mg, 8,053 100 16,410 100 9,159 100 9,965 100 14,725 100 4,043 100 924 100 48 165 472 switch to Allopurinol 100 mg (63,279) Female (19,526) 2,430 30.2 5,449 33.2 2,728 29.8 2,932 29.4 4,456 30.3 1,271 31.4 260 28.1 43 153 463 Male (43,747) 5,623 69.8 10,959 66.8 6,430 70.2 7,033 70.6 10,266 69.7 2,772 68.6 664 71.9 50 171 476 Other (6) 0 0 2 0 1 0 0 0 3 0 0 0 0 0 60 335 790 Allopurinol 300 mg, 1,563 100 6,577 100 2,850 100 2,403 100 2,756 100 571 100 101 100 31 97 285 switch to Febuxostat 40 mg (16,821) Female (5,503) 629 40.2 2,463 37.4 868 30.5 596 24.8 751 27.2 174 30.5 22 21.8 30 90 218 Male (11,318) 934 59.8 4,114 62.6 1,982 69.5 1,807 75.2 2,005 72.8 397 69.5 79 78.2 44 119 311 Other (0) 0 0 0 0 0 0 0 0 0 0 0 0 0 0 - - - Allopurinol 300 mg, 504 100 3,264 100 1,539 100 1,350 100 1,443 100 282 100 55 100 57 120 304 switch to Febuxostat 80 mg (8,437) Female (2,042) 148 29.4 916 28.1 345 22.4 253 18.7 302 20.9 64 22.7 14 25.5 34 90 264 Male (6,395) 356 70.6 2,348 71.9 1,194 77.6 1,097 81.3 1,141 79.1 218 77.3 41 74.5 60 123 318 Other (0) 0 0 0 0 0 0 0 0 0 0 0 0 0 0 - - -

163 of 169 Allopurinol (any 5,240 100 17,736 100 9,669 100 9,180 100 11,370 100 2,654 100 552 100 53 134 378 strength)3, switch to Febuxostat (any strength)2 (56,401) Female (20,659) 2,281 43.5 6,994 39.4 3,407 35.2 3,017 32.9 3,828 33.7 964 36.3 168 30.4 38 116 349 Male (35,740) 2,959 56.5 10,742 60.6 6,262 64.8 6,162 67.1 7,541 66.3 1,690 63.7 384 69.6 60 150 394 Other ( 2) 0 0 0 0 0 0 1 0 1 0 0 0 0 0 314 365 416

Table 4. Distribution of cumulative exposure duration, among patients with evidence of switch, by age, in the SDD, 2009-2016

<1 month 1 - <3 3 - <6 6 months - 1 - <3 years 3 - <5 years 5+ years Q1 Median Q3 months months <1 year

Exposures (n) N % N % N % N % N % N % N % days days days

Febuxostat 40 mg, 902 100 3,475 100 1,920 100 1,548 100 1,479 100 242 100 32 100 46 108 267 switch to Febuxostat 80 mg (9,598)

21-44 (963) 98 10.9 383 11 188 9.8 153 9.9 116 7.8 20 8.3 5 15.6 30 91 229

45-64 (2,808) 256 28.4 1,065 30.6 571 29.7 442 28.6 394 26.6 70 28.9 10 31.3 44 100 243

65+ (5,827) 548 60.8 2,027 58.3 1,161 60.5 953 61.6 969 65.5 152 62.8 17 53.1 50 115 281

Febuxostat 40 mg, 1,433 100 4,712 100 1,703 100 1,492 100 1,682 100 292 100 40 100 30 90 244 switch to Allopurinol 100 mg 11,354

21-44 (824) 74 5.2 431 9.1 114 6.7 85 5.7 97 5.8 22 7.5 1 2.5 30 60 181

164 of 169 45-64 (2,670) 301 21 1,167 24.8 434 25.5 335 22.5 353 21 69 23.6 11 27.5 30 86 222

65+ (7,860) 1,058 73.8 3,114 66.1 1,155 67.8 1,072 71.8 1,232 73.2 201 68.8 28 70 30 90 262

Febuxostat 40 mg, 723 100 3,359 100 1,201 100 961 100 1,023 100 174 100 23 100 30 90 219 switch to Allopurinol 300 mg (7,464)

21-44 (900) 92 12.7 458 13.6 132 11 108 11.2 93 9.1 13 7.5 4 17.4 30 60 179

45-64 (2,397) 207 28.6 1,145 34.1 391 32.6 305 31.7 287 28.1 54 31 8 34.8 30 80 201

65+ (4,167) 424 58.6 1,756 52.3 678 56.5 548 57 643 62.9 107 61.5 11 47.8 30 90 249

Febuxostat 80 mg, 132 100 581 100 358 100 348 100 468 100 90 100 17 100 60 173 433 switch to Febuxostat 40 mg (1,994)

21-44 (158) 8 6.1 48 8.3 36 10.1 27 7.8 32 6.8 6 6.7 1 5.9 60 149 358

45-64 (587) 46 34.8 188 32.4 102 28.5 103 29.6 124 26.5 15 16.7 9 52.9 60 148 369

65+ (1,249) 78 59.1 345 59.4 220 61.5 218 62.6 312 66.7 69 76.7 7 41.2 62 180 482

Febuxostat 80 mg, 164 100 772 100 297 100 271 100 291 100 34 100 7 100 30 90 264 switch to Allopurinol 100 mg (1,836)

21-44 (171) 11 6.7 81 10.5 36 12.1 19 7 22 7.6 2 5.9 0 0 30 73 192

45-64 (540) 42 25.6 247 32 77 25.9 77 28.4 83 28.5 10 29.4 4 57.1 30 90 259

65+ (1,125) 111 67.7 444 57.5 184 62 175 64.6 186 63.9 22 64.7 3 42.9 30 92 273

165 of 169 Febuxostat 80 mg, 173 100 894 100 326 100 289 100 331 100 47 100 6 100 30 90 270 switch to Allopurinol 300 mg

(2,066)

21-44 (298) 24 13.9 142 15.9 51 15.6 35 12.1 42 12.7 3 6.4 1 16.7 30 76 215

45-64 (760) 55 31.8 345 38.6 122 37.4 104 36 112 33.8 20 42.6 2 33.3 30 90 254

65+ (1,008) 94 54.3 407 45.5 153 46.9 150 51.9 177 53.5 24 51.1 3 50 30 92 290

Febuxostat (any 2,261 100 7,220 100 2,930 100 2,747 100 3,433 100 671 100 101 100 30 95 308 strength)2, switch to Allopurinol (any strength) (19,363)

21-44 (1,773) 177 7.8 789 10.9 266 9.1 216 7.9 267 7.8 49 7.3 9 8.9 30 78 246

45-64 (5,335) 537 23.8 2,110 29.2 845 28.8 746 27.2 869 25.3 192 28.6 36 35.6 30 92 290

65+ (12,255) 1,547 68.4 4,321 59.8 1,819 62.1 1,785 65 2,297 66.9 430 64.1 56 55.4 30 100 329

Allopurinol 100 22,217 100 46,871 100 26,567 100 22,379 100 22,660 100 4,364 100 765 100 32 100 272 mg, switch to Allopurinol 300 mg

(145,823)

21-44 (16,057) 2,503 11.3 5,929 12.6 3,085 11.6 2,367 10.6 1,860 8.2 258 5.9 55 7.2 30 90 209

45-64 (45,039) 6,687 30.1 15,071 32.2 8,434 31.7 6,919 30.9 6,567 29 1,145 26.2 216 28.2 30 97 254

65+ (84,727) 13,027 58.6 25,871 55.2 15,048 56.6 13,093 58.5 14,233 62.8 2,961 67.9 494 64.6 37 107 299

166 of 169 Allopurinol 100 3,785 100 11,962 100 5,562 100 4,786 100 4,820 100 924 100 183 100 30 93 270 mg, switch to Febuxostat 40 mg

(32,022)

21-44 (2,251) 249 6.6 959 8 414 7.4 334 7 254 5.3 32 3.5 9 4.9 30 90 205

45-64 (7,209) 781 20.6 2,849 23.8 1,337 24 1,072 22.4 972 20.2 162 17.5 36 19.7 30 90 238

65+ (22,562) 2,755 72.8 8,154 68.2 3,811 68.5 3,380 70.6 3,594 74.6 730 79 138 75.4 30 98 285

Allopurinol 100 726 100 4,390 100 1,921 100 1,594 100 1,453 100 200 100 27 100 30 92 245 mg, switch to Febuxostat 80 mg

21-44 (10,311) 68 9.4 536 12.2 197 10.3 147 9.2 111 7.6 11 5.5 2 7.4 30 82 188

45-64 (2,958) 201 27.7 1,285 29.3 579 30.1 469 29.4 369 25.4 49 24.5 6 22.2 30 90 220

65+ (6,281) 457 62.9 2,569 58.5 1,145 59.6 978 61.4 973 67 140 70 19 70.4 38 99 270

Allopurinol 300 8,053 100 16,410 100 9,159 100 9,965 100 14,725 100 4,043 100 924 100 48 165 472 mg, switch to Allopurinol 100 mg

(63,279)

21-44 (6,807) 840 10.4 1,901 11.6 1,110 12.1 1,227 12.3 1,360 9.2 305 7.5 64 6.9 32 135 369

45-64 (18,400) 2,364 29.4 4,834 29.5 2,803 30.6 2,932 29.4 4,139 28.1 1,022 25.3 306 33.1 39 152 450

65+ (38,072) 4,849 60.2 9,675 59 5,246 57.3 5,806 58.3 9,226 62.7 2,716 67.2 554 60 55 180 506

Allopurinol 300 1,563 100 6,577 100 2,850 100 2,403 100 2,756 100 571 100 101 100 31 97 285 mg, switch to

167 of 169 Febuxostat 40 mg (16,821)

21-44 106 6.8 735 11.2 342 12 299 12.4 303 11 32 5.6 7 6.9 41 104 270 (1,824)

45-64 (4,931) 456 29.2 1,860 28.3 884 31 761 31.7 781 28.3 152 26.6 37 36.6 30 100 281

65+ (10,066) 1,001 64 3,982 60.5 1,624 57 1,343 55.9 1,672 60.7 387 67.8 57 56.4 32 92 291

Allopurinol 300 504 100 3,264 100 1,539 100 1,350 100 1,443 100 282 100 55 100 57 120 304 mg, switch to Febuxostat 80 mg

(8,437)

21-44 (1,225) 43 8.5 473 14.5 230 14.9 228 16.9 212 14.7 30 10.6 9 16.4 60 120 299

45-64 (2,941) 166 32.9 1,143 35 545 35.4 487 36.1 490 34 93 33 17 30.9 57 116 298

65+ (4,271) 295 58.5 1,648 50.5 764 49.6 635 47 741 51.4 159 56.4 29 52.7 57 120 316

Allopurinol (any 5,240 100 17,736 100 9,669 100 9,180 100 11,370 100 2,654 100 552 100 53 134 378 strength), switch to Febuxostat (any strength ) (56,401)

21-44 (5,062) 312 6 1,713 9.7 929 9.6 891 9.7 1,019 9 153 5.8 45 8.2 60 136 348

45-64 (14,711) 1,210 23.1 4,692 26.5 2,666 27.6 2,509 27.3 2,865 25.2 619 23.3 150 27.2 56 133 362

65+ 3,718 71 11,331 63.9 6,074 62.8 5,780 63 7,486 65.8 1,882 70.9 357 64.7 51 134 392

168 of 169

APPEARS THIS WAY ON ORIGINAL

169 of 169