1 Arthritis Advisory Committee Meeting April 19th, 2001 NDA : GL701 for Systemic Lupus Erythematosus Genelabs Technologies, Incorporated

2 Presentation Outline

3 Consultants

4 GL701 (prasterone) Prasterone is the USAN designation for DHEA Prasterone is the synthetic equivalent of DHEA GL701 is the Genelabs formulation of Prasterone

5 Proposed Indications Improvement in SLE disease activity and/or symptoms in women with mild to moderate SLE Reduction in corticosteroid requirements in women with mild to moderate SLE

6 Background Robert Lahita, MD PhD

7 Systemic Lupus Erythematosus (SLE) Inflammatory autoimmune disease of unknown etiology Morbidity –Disease associated –Corticosteroid associated Corticosteroid use as high as 89% 1-2 Mortality 5-10% at 10 years –Early - active disease and infections –Late - atherosclerosis 1. Zonana-Nacach et al., Urowitz et al., ACR meeting 2000 (Abstract)

8 Damage within SLE SLICC/ACR Damage Index

9 DHEA and SLE: Preclinical Rationale Female NZB/W murine model –100% mortality at 10 months – Mortality reduced with DHEA administration Lucas et al., 1985; 2. Matsunaga et al., 1989; 3. van Vollenhoven and McDevitt, 1992 Murine in vitro studies –Altered cytokine profile with DHEA  IL-6,  IL Padgett and Loria, 1998; 5. Daynes et al.,

10 DHEA and SLE: Clinical Rationale Sex distribution in SLE, 90% F : 10% M Low levels of DHEA and other androgens in women with SLE Lahita et al., 1987; 2. Verthelyi et al., 2001 DHEA and testosterone further suppressed by corticosteroid use 3 3. Hedman et al., 1989 IL-2 levels suppressed in SLE 4-5 –In vitro (T lymphocytes) DHEA increased IL-2 production 6 DHEA inhibits IL-6 secretion (mononuclear cells) 7 4. Alcocer-Varela and Alarcon-Segovia, 1982; 5. Linker-Israeli et al., 1983; 6. Suzuki et al., 1991; 7. Straub et al.,

11 Study GL95-02 Baseline Endogenous DHEA-S and Testosterone Levels with and without Corticosteroid Treatment Baseline TestosteroneBaseline DHEA-S

12 Rationale for Androgen Therapy of SLE Endocrinologic: –Low androgen levels in women with lupus. –Higher oxidation of testosterone at C17 in women with lupus Immunologic: –Decrease of IL 4, IL 5, IL 6 (TH2) cytokines and increase of IL2 (TH2)

13 Efficacy Michelle Petri, MD

14 Stanford University Phase I/II Studies DHEA use in Women with SLE Double-Blind, Placebo Controlled Study (1) –28 women with mild to moderate SLE treated for 3 months SLE Disease Activity Index (SLEDAI), Physician Visual Analog Scale (VAS) stabilized or improved Patient VAS improved significantly (P = 0.022) Number of disease flares decreased (P = 0.053) Decreased prednisone requirements Open-Label Study (2) –50 women with SLE –Improvements similar to those noted in the placebo controlled study 1. van Vollenhoven et al., 1995 ; 2. van Vollenhoven et al.,

15 Overview of Clinical Trial Design Process Collaborative process between Genelabs, FDA and Consultants 1995 Arthritis Advisory Committee –Two efficacy per-patient endpoints steroid reduction improved disease activity 1999 Arthritis Advisory Committee – Clinical trial endpoints discussed

16 Burden of Disease Most patients have either recurrent flares or continuously active disease 1 Flares remain common in established disease 2 Morbidity also associated with corticosteroid use 3 1. Barr et al., 1999 ; 2. Petri et al., 1991; 3. Zonana-Nacach, et al.,

17 Clinical Domains of SLE Disease Activity SLEDAI SLAM Organ Damage Clinical Deterioration SLICC Damage Index Health Related Quality of Life KFSS Patient VAS SF

18 Development of Efficacy Endpoints for GL701 Clinical Trials 1) Reduction in Corticosteroid Requirements If SLEDAI was stable or improved, an algorithm dictated steroid taper 2) Improvement or Stabilization of SLE Based upon improvement or stabilization in each of SLEDAI, SLAM, KFSS and Patient VAS, without clinical deterioration

19 GL701 Principal SLE Clinical Trials

20 Study GL94-01 Objective Reduction in Corticosteroid Requirements

21 GL94-01: Corticosteroid Reduction Study Design Double-blind, randomized, parallel design GL or 200 mg/day vs. placebo 7-9 months dosing, assessments monthly Prednisone dose reduced at each visit if SLEDAI stable or improved, based on a pre-determined algorithm

22 GL94-01: Corticosteroid Reduction Entry Criteria Women with Mild to Moderate Steroid Dependent SLE Defined as: –Stable prednisone dose at entry mg/day and –Unsuccessful prednisone taper, or no taper, stable dose in last 12 weeks

23 GL94-01: Corticosteroid Reduction Primary Efficacy Endpoint (Responder) Sustained prednisone reduction : –Prednisone decreased to  7.5 mg/day For  2 consecutive months Including last visit

24 GL94-01: Corticosteroid Reduction Baseline Demographics

25 GL94-01: Corticosteroid Reduction Baseline Characteristics

26 GL94-01: Corticosteroid Reduction Impact of Baseline SLEDAI At the pre-study investigator meeting there was concern whether patients with 0 or low SLEDAI scores should be enrolled –Indicative of either smoldering disease that would flare when prednisone was tapered? or Indicative of inactive disease that was not steroid- dependent? To address this, a blinded analysis of responders, without treatment group attribution, was reviewed prior to study unblinding

27 GL94-01: Corticosteroid Reduction Blinded Analysis of Patients by Baseline SLEDAI N=28 N=26 N=42 N=53 N=

28 Study GL94-01: Corticosteroid Reduction Analysis of the 54 Patients with Baseline SLEDAI Scores of (51%) with score of 0 20 (38%) with scores of 2 due to serologies 6 (11%) with scores of 1-2 due to other: Mucosal ulcers (N = 2) Leukopenia (N = 1) Alopecia (N = 1) New rash or pleurisy (N = 2) Therefore, the SLEDAI 0-2 subgroup differed in clinical characteristics, not just in response

29 GL94-01: Corticosteroid Reduction Impact of Baseline SLEDAI Score (cont) These data suggested that baseline SLEDAI >2 (yes/no) might represent different populations Therefore, patients with baseline SLEDAI > 2 were defined as a subgroup prior to unblinding

30 GL94-01: Corticosteroid Reduction Patient Disposition

31 GL94-01: Corticosteroid Reduction Responders 26/64 28/63 35/ mg vs. placebo: ^P = 0.110*P = /45 18/47 23/

32 Study GL94-01 Responders /19 10/16 12/19 5/15 8/18 5/9 4/13 7/16 12/24 2/9 3/6 2/8 1/4 3/6

33 Responder Rate by Treatment and by Prednisone Dose There was a statistically significant (P = 0.039) difference in prednisone at baseline for the SLEDAI > 2 group between GL mg and placebo

34 GL94-01: Corticosteroid Reduction Mean Prednisone Reduction at Last Visit This endpoint does not fully reflect prednisone reduction because: –There was no algorithm for prednisone increases, and –This analysis only reflected prednisone reduction on the last day

35 GL94-01: Corticosteroid Reduction Individual Patient Example of Prednisone Dose Changes from Baseline to Last Visit

36 GL94-01: Corticosteroid Reduction Number of Days Prednisone  7.5 mg/day * Parametric: GL mg vs. placebo, P = ** Nonparametric: GL mg vs. placebo, P = by Wilcoxon Rank-Sum test ^ Nonparametric GL mg vs. placebo, P = by Wilcoxon Rank-Sum test

37 GL94-01: Corticosteroid Reduction Efficacy Summary All patients: –Sustained corticosteroid reduction (responder): 200 mg (55%) vs. placebo (41%), P = –Greater number of days on prednisone  7.5 mg/day (P = 0.069) In patients with baseline SLEDAI > 2: –Higher response rate: 200 mg (51%) vs. placebo (29%), P = –Dose response (test for trend, P = 0.033) –Greater number of days on prednisone  7.5 mg/day (P = 0.015)

38 Study GL95-02 Objective Improvement or stabilization in SLE

39 GL95-02: Improvement in SLE Study Design Double-blind, randomized, parallel design 12 month study; assessments every 90 days GL mg/day vs. placebo Concomitant prednisone, immunosuppressives, and antimalarials allowed at baseline and continued unchanged DEXA for BMD performed at 8 investigator sites for patients on chronic steroids prior to and during study

40 GL95-02: Improvement in SLE Entry Criteria Women with active SLE –SLAM  7 at screen and qualifying visits –Prednisone  10 mg/day –There was an evidence-based (GL94-01) protocol amendment to require active SLE (SLEDAI > 2) at baseline and enrollment increased to capture more of these patients

41 GL95-02: Improvement in SLE Primary Endpoint: Responder Improvement or stabilization in all of the following: –Two disease activity measures: SLEDAI and SLAM –Two constitutional measures: Patient VAS and KFSS Based on mean of on-treatment visits, compared to baseline mean and No clinical deterioration

42 GL95-02: Improvement in SLE Primary Endpoint: Responder (cont) Clinical Deterioration defined as: –New or progressive organ disease –Serious drug toxicity –New or increased dose of prednisone or cytotoxic agents

43 GL95-02: Improvement in SLE Development of the Analysis Plan No previous experience Collaborative process with FDA and consultants on study design Two additional key issues identified from inception of study to completion of the final analysis plan (Dec 95 - April 99) –Defining stabilization as part of responder definition (“window concept”) –Identifying primary analysis data set

44 GL95-02: Improvement in SLE Defining Stabilization for Each Instrument: Concept of a “Window” Two baseline pre-treatment evaluations of disease activity typically used in rheumatology clinical trials because of inherent variability in instruments Test/re-test variability of the instruments used in this trial is well known: Liang et al, 1989; Bombardier et al, 1992; Petri et al, 1992; DeLoach et al, 1998; Fitzgerald and Grossman, 1999 Definition of “stabilization” was not finalized prior to initiating the study

45 GL95-02: Improvement in SLE Evidence-based Confirmation of Pre-Defined “Window” Pre-defined window (Oct 1998): 0.5 SLEDAI and KFSS, 1.0 SLAM and 10 Patient VAS After study completion, a comparison of the differences for each patient in the two baseline visits (screening and qualifying visits < 10 days apart) showed the following variability: This agrees well with the pre-defined window

46 Example of a Patient Classified as a Responder when Using the “Window”

47 GL95-02: Improvement in SLE Secondary Endpoints Mean changes in scoring instruments: SLEDAI, SLAM, Patient VAS, KFSS Bone mineral density (BMD) by DEXA in patients on chronic corticosteroids Proportion of patients with SLE flare

48 GL95-02: Improvement in SLE Baseline Demographics: All Randomized (ITT)

49 GL95-02: Improvement in SLE Baseline Characteristics: All Randomized (ITT)

50 GL95-02: Improvement in SLE Patient Disposition: All Randomized (ITT)

51 GL95-02: Improvement in SLE Percent Responders: ITT P = /192 58/18965/146 86/147 * P = 0.017

52 GL95-02: Improvement in SLE Populations for Analysis Original protocol specified ITT Analysis plan specified per-protocol population (N = 346) –Patients treated for  60 days and at least 1 assessment beyond 60 days –Patients excluded from per protocol Excluded 32 patients with no post-baseline measurements Excluded 1 placebo patient who took no drug for first 90 days (non- responder) Excluded 2 placebo patients with < 60 days treatment (1 responder, 1 non-responder) –This population is virtually identical to a modified ITT (3 patient difference)

53 GL95-02: Improvement in SLE Reasons for Withdrawal for Patients Excluded from Per-Protocol Population Excluded Patients N = 35 Possibly Related AE’s GL701 N = 4 Placebo N = 3 Lack of Efficacy GL701 N = 4 Placebo N = 2 Unrelated to Safety or Efficacy GL701N = 8 Placebo N = 7 No Information GL701 N = 3 Placebo N =

54 GL95-02: Improvement in SLE Baseline Characteristics of Excluded Patients from Per-Protocol Analysis

55 GL95-02: Improvement in SLE Percent Responders: Per-Protocol (With Window) * P = /17699/17065/133 87/132 ** P =

56 Study GL95-02 Responders by Baseline SLEDAI Score (Per-protocol)* /43 12/38 20/36 15/25 35/62 43/58 5/22 27/32 5/13 8/12 *With Window

57 GL95-02: Improvement in SLE Patients with Baseline SLEDAI Scores 0-2 Are a Different Population (Inactive Disease) Baseline SLEDAI ScoreStudy GL94-01Study GL (51%)38 (43%) 1-2 due to serologies20 (38%)25 (28%) 1-2 due to other 6 (11%)25 (28%)

58 GL95-02: Improvement in SLE Mean Changes in Scoring Instruments* (Per-protocol) *Baseline SLEDAI >

59 GL95-02: Improvement in SLE Flare Definition* Any one of the following: *Definition of definite flare adapted from SELENA (NIH) study

60 GL95-02: Improvement in SLE Percent of Patients with Flares P = ns 47/17637/17041/133 31/

61 Study GL95-02 Selected Baseline Characteristics of Patients Assessed for BMD

62 GL95-02: Improvement in SLE Percent Change in BMD at 1-year (N = 37) *P = 0.004**P = ***

63 GL95-02: Improvement in SLE Percent of Patients with >3% Gain or Loss in BMD at 1 Year

64 GL95-02: Improvement in SLE Efficacy Summary ITT: –Baseline SLEDAI > 2 group: higher responder rate in GL701 (59%) vs. placebo (45%), P=0.017 (with window) Per-protocol: –Higher responder rate in GL701 (58%) vs. placebo (46%), P=0.018 –Baseline SLEDAI > 2 group: higher responder rate in GL701 (66%) vs. placebo (49%), P=0.005 Secondary: –Improved BMD (L-Spine, P=0.004 vs. placebo) in patients on chronic corticosteroids –Patient global assessment improved –Flares reduced

65 GBL96-01: Improvement in SLE – Taiwan Study Design Double-blind, randomized, parallel design Objective: disease improvement (similar to GL95-02) Women with active SLE –Baseline SLAM  7 and later amended to also require SLEDAI > 2 required for entry GL mg vs. placebo 6 month study duration

66 GBL96-01: Taiwan Study Demographic & Baseline Characteristics

67 GBL96-01: Taiwan Efficacy Results: ITT

68 GBL96-01: Taiwan Time to First Flare GL701 Placebo Days Percent without Flares P = 0.044

69 Overall Efficacy Summary Disease Activity –Improvement and stabilization in SLE activity –Fewer patients with disease flares Damage –Sustained reduction of corticosteroids –Improvement in bone mineral density in prednisone- treated patients Health Related Quality of Life –Improvement in patient global assessment (VAS)

70 Statistical Discussion Frank Hurley, PhD

71 Statistical Issues Strategy of new drug development in uncharted territory Target Population: Predefined Subgroup analysis based on baseline SLEDAI>2 Measurement tolerance for definition of stabilization of disease Differential outcomes for two primary endpoints for study GL94-01 All randomized ITT vs. modified ITT vs. per- protocol

72 Strategy in Uncharted Territory Flexibility needed in design and analysis of clinical trials in diseases such as SLE Flexible approach with careful planning, proper execution and scientific rigor would not compromise scientific validity –Target population (SLEDAI >2, patients with active disease) based on GL94-01 and implemented in an amendment in GL95-02 –Per protocol population Minimizing noise and maximizing ability to detect treatment difference; a strategy needed without prior knowledge of treatment effect using an instrument (responder) with unknown properties –ITT population Preferred with prior knowledge of treatment effect in target population and measurement instrument sensitivity - allows sample size calculation with adequate statistical power

73 Target Population: Predefined Subgroup Analysis Based on SLEDAI>2 Baseline SLEDAI > 2 identified as clinically important subgroup based on blinded review of aggregated data in study GL94-01 Prednisone target reduction achieved in 2/3 of subjects with baseline SLEDAI  2 regardless of treatment group Analysis of GL94-01 shows significantly different responses for the subgroups (placebo group: SLEDAI  2: 68% responders; SLEDAI>2: 29% responders) SLEDAI >2 defined as protocol inclusion target population criterion (by amendment) for study GL95-02 The appropriateness of the target population definition was confirmed in GL95-02 study

74 Measurement Tolerance for Definition of Stabilization of Disease All of the scales used to assess efficacy in GL95-02 have inherent intra-patient, intra-rater variability (test-retest variability) Definition of stabilization should include reasonable tolerance to inherent measurement variability –Example: ACR20 for improvement in rheumatoid arthritis trials –Too stringent if no variability allowed in all four components Tolerance window concept discussed during early stages of study; proposal finalized prior to breaking blind

75 Sensitivity Analysis - Per Patient Window (% Change from Baseline) Response Rate by Window Size Target Population Baseline SLEDAI > 2, GL95-02 Windows in this area require improvement in all 4 instruments (SLAM, SLEDAI, Patient VAS, KFSS) A too wide window turns the majority into responders. This is not Clinically Meaningful

76 Measurement Tolerance for Definition of Stabilization of Disease Robustness of pre-defined window assessed using % of baseline score on per patient basis –Conclusions unchanged for a range of windows from - 3% to -30% Placebo response of 30% to 45% is not uncommon in mild to moderate diseases (esp. rheumatologic diseases), particularly with significant background therapy

77 GL94-01: Corticosteroid Reduction Differential Outcomes for Two Primary Endpoints Protocol Specified two primary endpoints: –Responder: decrease in prednisone dose to  7.5 mg/day for three consecutive visits (i.e. 2 months) including last visit (Subpart E) –Percent decrease in prednisone dose at last visit compared to baseline Responder endpoint based on down titration of dose to pre-specified lower limit For the target population (SLEDAI>2), the GL mg group had a 51% responder rate compared to 29% for the placebo group (p=0.031)

78 GL94-01: Corticosteroid Reduction Differential Outcomes for Two Primary Endpoints Percent reduction in dose highly influenced by large percentage dose increases in a small number of patients The average percent reduction from baseline to the last visit for the overall population was 30% for the GL mg compared to 35% for the placebo group Seven patients increased % of baseline; excluding these data results in mean percent reduction of 48% for GL mg compared to 41% for placebo

79 GL95-02: Improvement in SLE Sensitivity Analysis ITT Subset (Baseline SLEDAI > 2) with Window Patients had no post baseline assessments but: –Reported deterioration, or –Were discontinued early due to lack of efficacy (These patients are reclassifed as non-responders) Results: –GL mg Responders 58% –Placebo Responders 43% P =

80 GL95-02: Improvement in SLE All Randomized ITT vs. Modified ITT vs. Per- Protocol Population All Randomized ITT –Patients without post-baseline measurements were considered as non-responders: Patients without treatment Missing all post baseline measurements No evidence of clinical deterioration Modified ITT –Commonly used in clinical trials –Patients excluded if no post baseline assessments and no known clinical deterioration Per Protocol Population –Similar to modified ITT population, excludes 3 more patients: 2 for less then 60 days of treatment, 1 for a major protocol violation

81 GL95-02: Improvement in SLE All Randomized ITT vs. Modified ITT vs. Per- Protocol Population (Continued) Results of Modified ITT and per-protocol results closely similar No apparent bias observed using either population No evidence that excluded patients are non- random: test of null-hypothesis is valid

82 Conclusion For the target population (SLEDAI>2) using the defined windows for stabilization, all analyses show highly significant responder rates for GL mg compared to placebo Use of target population and tolerance window are matters of clinical judgment, not statistical principle

83 Safety Michelle Petri, MD

84 Presentation of Safety Data

85 Duration of Exposure to GL

86 All Reported Deaths: GL701 Group (N = 495)

87 All Reported Deaths: Placebo Patients who Never Received GL701 (N = 77)

88 Reported Serious Adverse Events from Studies GL94-01 and GL

89 Studies GL94-01 & GL95-02 Withdrawals Due to Medically Serious Adverse Events GL701 Hepatitis C GI bleed Psychosis Pulmonary edema Renal deterioration Placebo 2 Septicemia Hepatitis Suicidal depression Pneumonia Coronary artery spasm Carcinoma of the lung Pseudotumor cerebri

90 Premature Withdrawals: Total and Due to Androgenic Complaints from Studies GL95-02 and GL

91 Adverse Events with a Frequency of  10%

92 Selected Adverse Events with a Frequency of <10%**

93 Mean Testosterone Pre and Post Treatment Pooled Data GL94-01 and GL95-02 *Williams Textbook of Endocrinology,

94 Mean Estradiol Pre and Post Treatment Pooled Data GL94-01 and GL95-02 Pre-menopausal Patients N = 101N = 41N =

95 Mean / Median Estradiol Pre and Post Treatment Pooled Data GL94-01 and GL95-02 Post-menopausal Women* Not on Hormone Replacement Therapy * Defined as women with estradiol levels of < 25 pg/ml MedianNormal Postmenopausal

96 Mean / Median Estradiol Pre and Post Treatment Pooled Data GL94-01 and GL95-02 Post-menopausal Women on Hormone Replacement Therapy Median Normal Post- Menopausal

97 Breast Cancer

98 Breast Cancer Incidence* Normalized for Period of Observation for all Patients and for those  45 years of age *Control rates (no HRT) reported as 3.9 per 1000 women years and unopposed estrogen use rates reported 4.4 per 1000 women years 1 1. Schairer et al.,

99 Implications: Effects on Hormones Testosterone levels increased –Androgenic effects observed were acne and hirsutism Estradiol levels increased in post-menopausal women –Increases consistent with those reported for hormone replacement therapy –No increase in breast carcinoma – one to two years observation –No significant increase in vaginal bleeding –No endometrial hyperplasia observed Increased bone mineral density

100 Routine Clinical Laboratories No evidence of significant effects on: –CBC –Liver function tests –BUN and Creatinine –Routine serum chemistries

101 Mean Changes in Serum Lipids

102 HDL-Cholesterol Change

103 Possible Mechanisms of Decrease in HDL and Triglycerides Testosterone increases hepatic lipase activity Increased hepatic lipase activity enhances HDL clearance, and possibly affects reverse cholesterol transport (removal of cholesterol from tissues) Experimental evidence suggests an increase in hepatic lipase activity may be anti-atherogenic Rabbit studies with DHEA indicate anti- atherogenic effects 1-4 ; mechanism unknown 1. Alexandersen et al., 1999; 2. Arad et al.,1989; 3. Eich et al., 1993; 4. Gordon et al.,

104 Serum Complement

105 Study GL96-02 Percent Change in Serum C3 Complement in Normal Premenopausal Women (N=14) Mean-2.3% Median-1.4%

106 Studies GL96-02 and GL94-01: Mean C3 Complement Percent Change from Baseline in Normal Women and SLE Patients During GL701 Treatment PlaceboGL mg GL mg PlaceboGL mg GL mg GL mg Study GL96-02 Normal Female Volunteers Study GL94-01 SLE Patients

107 Shift Table: Change in C3 from Baseline to Last Visit

108 Patients with C3 Normal to Low Of 14 placebo with decrease in C3: –2 had isolated new onset hematuria Of 36 GL mg with decrease in C3: –3 had isolated new onset hematuria –2 had isolated increased proteinuria 2592  24,062 mg/24 hr 2420  4648 mg/24 hr –2 had increase in serum creatinine (  0.3 mg/dl) 2.2  2.6 mg/dl 4.2  5.2 mg/dl None of these patients (placebo or GL mg) received immunosuppressive therapy for renal lupus

109 Possible Mechanism of Action: Effects on C3 Complement Mechanism may be decreased production rather than increased consumption –DHEA decreases IL-6 production, which may mediate hepatic complement synthesis –DHEA may decrease hepatic production of some proteins including complement –Testosterone therapy for Klinefelter’s has been shown to decrease serum complement - no autoimmune manifestations 1 1. Yesilova, et al

110 Implications: Effects on C3 Decrease in C3 did not appear to correlate with increased disease activity Appears not to be associated with worsening renal disease

111 Patients with Hematuria as an Adverse Event

112 Patients with Creatinine Increase of  0.3 mg/dl from Baseline to Last Visit

113 Patients with Clinically Meaningful Increases in 24 Hour Urine Protein at Last Visit

Hour Urine Protein at Last Visit: Patients with Doubling of Protein for at least 2 Visits GL701 N = 23 Placebo N = 14 Normal GL701 N = 7 Placebo N = 2 Modest (300-1,000) GL701 N = 5* Placebo N = 4 Mild (<300) GL701 N = 11 Placebo N = 8 Moderate (>1,000) GL701 N = 0 Placebo N = 0 Normal at Baseline *348, 424, 303, 404, 325

115 Study GL94-01 Renal Analysis per Patients Identified by FDA Reviewer Any Two Abnormalities* * C3 and/or C4 counted as one Decreased CCr: from normal to abnormal; any decrease from abnormal BL Increased Total proteinuria: from none to  500; from abnormal to  1000 Rbcs: from 0 to  10; from abnormal at BL to  25 Complement 3: from normal at BL to < lower limits of normal

116 Study GL95-02 Renal Analysis per Patients Identified by FDA Reviewer Any Two Abnormalities* * C3 and/or C4 counted as one Decreased CCr: from normal to abnormal; any decrease from abnormal BL Increased Total proteinuria: from none to  500; from abnormal to  1000 Rbcs: from 0 to  10; from abnormal at BL to  25 Complement 3: from normal at BL to < lower limits of normal

117 Signs of Renal Flare

118 Signs of Renal Flare Medical Reviewer’s definition of a renal signal, except decrease in C3, C4, or increase in dsDNA counted only once.

119 DHEA Administration in Severe SLE 1 Responders at 6 Months Responder definition –Creatinine clearance stable –>50% reduction proteinuria –Inactive urinary sediment Responders (patients with nephritis): –DHEA 6/8 –Placebo 4/6 1. van Vollenhoven et al,

120 Implications: Renal Safety Signal for renal safety in GL94-01 is not confirmed in GL95-02 The reduction in C3 appears to be a marker of reduced hepatic synthesis, not a renal safety signal Androgens may increase renal plasma flow but do not cause glomerular hypertension. This may explain the few patients who had mild to modest increases in proteinuria on GL701, without overt evidence of nephritis. DHEA administration to severely ill lupus patients with nephritis led to improvement in 6 out of 8, with none worsening

121 Overall Safety Summary Majority of adverse events are androgenic (acne and hirsutism) and only led to a small number of withdrawals Clinical laboratory changes reflect known hormonal effects, primarily androgenic –Increase in testosterone –Decrease in triglycerides, HDL-C –Increase in estradiol Decrease in C3, without adverse clinical consequences Modest increase in proteinuria observed in some GL701- treated patients, but without any signal of renal flares, including decreased creatinine clearance

122 Clinical Perspective Murray Urowitz, MD

123 Review of GL701 Clinical Designs and Outcomes Rationale Challenges in study design Advantage Outcomes: Important Findings

124 GL94-01: Corticosteroid Reduction Rationale –Large body of patients are on long-term use of steroids to control disease activity This contributes to additional, and potentially significant, damage in SLE Challenges in study design –Forced titration of steroids inherently difficult –Other efficacy variables expected to remain stable –Some patients, assumed to be steroid dependent, are mistakenly kept on unnecessarily high doses of steroids: this affected the entry criteria for steroid dependence

125 GL94-01: Corticosteroid Reduction Advantage –Clinically meaningful and practical objective Outcomes: Important Findings –Correlation between disease activity and steroid dependency Most patients with SLEDAI  2 were clinically inactive and capable of prednisone reduction –Greater treatment effect (i.e., steroid reduction) demonstrated in SLEDAI > 2 subgroup –Treatment effect was present in those receiving high doses [>15-30 mg/d] as well as low doses [10-15 mg/d] of prednisone at study entry

126 GL95-02: Improvement in SLE Rationale for Study Design –A significant number of SLE patients flare or deteriorate over a 1-year period Challenges –Defining an appropriate endpoint Incorporating multiple instruments into one composite “responder” endpoint Characterizing stable disease as a responder endpoint: “Window” concept developed to account for instrument variability Responder definition previously untested in lupus clinical trials

127 GL95-02: Improvement in SLE (cont) Challenges (cont) –Identifying the patient population that could be treated over 1-year Mild to Moderate SLE Patients needed to be on stable doses of steroids, yet active Maintaining patients in study for 1-year

128 GL95-02: Improvement in SLE Advantage –Responder integrated 3 domains of SLE over a 1-year period to fully assess the impact on lupus –Deterioration was captured at any time during study, not just at the pre-determined visits Outcomes –“Responder” composite endpoint proved successful –Use of flare to assess efficacy at any time point –Importance of identifying active SLE patient population (i.e.,SLEDAI > 2) confirmed

129 Potential Role of GL701 in the Management of SLE Patients For Mild to Moderate Lupus –Effective in controlling disease manifestations –Positive impact on health related quality of life Patient self assessment of global impact of disease Ability to allow withdrawal of steroids while preventing worsening of disease activity –Benefits greater than risks Benefit has been demonstrated for all disease domains Possible long term benefits include improved BMD Important immediate risks not identified in clinical trials Possibility of synergy with other lupus drugs Place in Lupus therapy armamentarium