229
`
`Fulvestrant versus Anastrozole for the Treatment of
`
`Advanced Breast Carcinoma in Postmenopausal
`Women
`
`A Prospective Combined Analysis of Two Multioenter Trials
`
`John F. R. Robertson, Mp.‘
`C. Kent Osborne, M.D.2
`Anthony Howell, rvr.o.3
`Stephen E. Jones, M.D.4’5
`Louis Mauriac, M.D.°
`Matthew Ellis, M.D., Ph.D.7
`Ulrich R. Kleeberg, M.D., mo.“
`Steven E. Come, m.n.9
`Ignace Vergote, M.D., pn.n.‘°
`Stan Gertler, Mp.“
`Aman Buzdar, M.D.12
`Alan Webster, vrsc.”
`Charles Morris, M.B., ch.n.‘3
`
`‘ Department of Surgery, Nottingham City Hospital,
`Nottingham, United Kingdom.
`
`2 Breast Center at Baylor College of Medicine and
`The Methodist Hospital, Houston, Texas.
`
`3 Cancer Research UK (CRUK) Department of Med-
`ical Oncology, Christie Hospital, Manchester,
`United Kingdom.
`
`4 Bay|or—Sammons Cancer Center, Dallas, Texas.
`
`5 US Oncology Research, Houston, Texas.
`
`5 Bergonie Institute, Bordeaux, France.
`
`7 Lombardi Cancer Center, Washington, DC.
`
`8 Haematologische/Oncologische Praxis, Hamburg,
`Germany.
`
`9 Beth Israel Deaconess Medical Center, Boston,
`Massachusetts.
`
`‘O Department of Gynaecology and Gynaecological
`Oncology, University Hospital of Leuven, Leuven,
`Belgium.
`
`‘‘ Ottawa Regional Cancer Center, Ottawa, On-
`tario, Canada.
`
`‘2Breast Medical Oncology, The University of
`Texas M. D. Anderson Cancer Center, Houston,
`Texas.
`
`© 2003 American Cancer Society
`DOI 10.1002/cncr.11468
`
`BACKGROUND. Fulvestrant (ICI 182,780) is a new type of estrogen receptor (ER)
`antagonist that down—regulates the ER and has no known agonist effects. The
`authors report the prospectively planned combined analysis of data from 2 Phase
`Ill trials comparing fulvestrant 250 mg monthly (7; = 428) and anastrozole 1 mg
`daily (n = 423) in postmenopausal women with advanced breast carcinoma (ABC)
`who previously had progressed after receiving endocrine treatment.
`METHODS. The primary endpoint was time to progression (TTP). Secondary end-
`points included objective response (OR), duration of response (DOR), and tolera-
`bility. The trials were designed to demonstrate superiority of fulvestrant over
`anastrozole. Noninferiority of fulvestrant versus anastrozole was determined using
`a retrospectively applied statistical test
`RESULTS. At a median follow—up of 15.1 months, ~ 83% of patients in each
`treatment arm had progressed. The median TTP was 5.5 months in the fulvestrant
`group and 4.1 months in the anastrozole group, and the OR rates were 19.2% and
`16.5% for fulvestrant and anastrozole, respectively (although the difference be-
`tween treatments was not statistically significant). In patients who responded,
`further follow—up (median, 22.1 months) was performed to obtain more complete
`information on DOR; the median DOR (from randomization to disease progres-
`sion) in patients who responded to treatment was 16.7 months in the fulvestrant
`group and 13.7 months in the anastrozole group. In a statistical analysis of DOR
`(using all randomized patients; from the start of response to disease progression),
`DOR was significantly longer for patients in the fulvestrant group compared with
`patients in the anastrozole group. Both drugs were tolerated well; withdrawals due
`to drug—related adverse events were 0.9% and 1.2% in the fulvestrant group and the
`
`‘3AstraZeneca
`United Kingdom.
`
`Pharmaceuticals, Macclesfield,
`
`Supported by a grant from AstraZeneca Pharma-
`ceuticals.
`
`Dr. Robertson has received research grants and
`speaker's honoraria from Astrazeneca and Novar-
`tis, and he has served on the advisory board of
`AstraZeneca. Dr. Howell has received speaker's
`honoraria from and served on the advisory board
`of AstraZeneca. Dr. Ellis has received research
`grants from Genentech and Novartis and speaker's
`honoraria from AstraZeneca, Genentech, and No-
`vartis; and he has performed consulting work for
`Johnson & Johnson and Novartis. Dr. Come has
`
`performed consulting work for and received
`speaker's honoraria from Astrazeneca and Novar-
`tis. Dr. Buzdar has received research grants from
`AstraZeneca, Bristol—Myers Squibb, Chugai, Eli
`Lilly, Genentech, Pharmacia, and Taiho Pharma-
`ceuticals. Dr. Webster and Dr. Morris are employ-
`ees of AstraZeneca.
`
`Address for reprints: John F. R. Robertson, M.D.,
`Department of Surgery, Nottingham City Hospital,
`Hucknall Road, Nottingham, NG5 1PB, UK; Fax:
`(011) 44(0) 115-840-2618; E—rnail: john.robertson@
`nottingham.ac.uk.
`
`revision received
`Received January 29, 2003;
`March 24, 2003; accepted March 24, 2003.
`
`Astrazeneca Ex. 2061 p. 1
`Mylan Pharrns. Inc. v. Astrazeneca AB IPR2016-01324
`
`

`
`230
`
`CANCER July 15, 2003 / Volume 98 I Number 2
`
`anastrozole group, respectively. The incidence of joint disorders was significantly
`lower in the fulvestrant group (P = 0.0036).
`CONCLUSIONS. Fulvestrant was tolerated well and was at least as effective as
`
`anastrozole in the second-line treatment of patients with ABC. This new hormon-
`altherapy may provide a valuable treatment option for ABC in postmenopausal
`women. Cancer 2003;963:229 -38. © 2003 American Cancer Society.
`
`KEYWORDS: fulvestrant, estrogen receptor antagonist, duration of response, com-
`bined analysis.
`
`he most widely used hormonal treatment for pa-
`tients with breast carcinoma is the antiestrogen
`tamoxifen. Most patients with hormone-sensitive
`breast carcinoma currently receive tamoxifen at some
`stage during their treatment. Many of these patients
`eventually develop tamoxifen-resistant disease, which
`leaves clinicians with the problem of how best to
`manage patients with hormone-sensitive tumors.
`Other hormonal therapies include the selective aro-
`matase inhibitors (AIS),
`including anastrozole and
`letrozole, and the steroidal agent exemestane. Both
`anastrozole” and letrozole3 are effective and well tol-
`
`erated. Fulvestrant (FaslodexTM) is a new estrogen re-
`ceptor (ER) antagonist that, unlike tamoxifen, is de-
`void of agonist activity.4 Binding of fulvestrant to the
`ER induces a rapid loss of ER protein from breast
`carcinoma cells.5 Fulvestrant down-regulates the ER
`in a dose dependent manner, as indicated by a dose-
`related reduction in the ER index.6 Compared with
`tamoxifen,
`fulvestrant consistently reduces tumor
`progesterone receptor
`(PgR) contents This novel
`mode of action distinguishes fulvestrant from all other
`antiestrogens currently in clinical use (eg, tamoxifen,
`toremifene, and raloxifene).
`In preclinical studies, fulvestrant was markedly
`more effective than tamoxifen at inhibiting the growth
`of human breast carcinoma cells in vitro.7 Further-
`
`more, fulvestrant was effecfive against tarnoxifen-re-
`sistant breast carcinoma xenographs in an in vivo
`mouse model.” Phase I clinical trials demonstrated
`
`that a short-acting formulation of fulvestrant admin-
`istered daily for 7 days before primary breast surgery
`was tolerated well and had antiestrogenic and antipro-
`liferative effects,9 whereas a Phase II study with the
`current long-acting formulation, which was adminis-
`tered once monthly, showed that fulvestrant was ef-
`fective in women who had breast carcinoma that pro-
`gressed after tamoxifen therapy.“””
`The current article reports the combined analysis
`of 2 Phase III clinical trials (Trial 0020 and Trial 0021),
`each of which compared a once-monthly intramuscu-
`lar (i.m.) injection of fulvestrant 250 mg with a once-
`daily oral dose of the third-generation, nonsteroidal
`AI, anastrozole 1 mg. Both were multicenter, random-
`
`ized, controlled, parallel-group trials. Each trial com-
`pared the efficacy and tolerability of fulvestrant and
`anastrozole in postmenopausal women with advanced
`breast carcinoma (ABC) who previously had disease
`progression after receiving endocrine treatment. The
`results for the individual trials have been reported
`previously.13’” Statistical plans included prospectively
`designed analyses of the combined data that are pre-
`sented in the current report.
`
`MATERIALS AND METHODS
`Data were combined from 2 trials (Trial 0020 and Trial
`0021) comparing the efficacy and tolerability of fulves-
`trant 250 mg given by intramuscular (i.m.) injection
`once monthly with anastrozole 1 mg given orally once
`daily. Trial 0020 was an open-label, randomized, mul-
`ticenter, parallel-group trial conducted in Europe,
`Australia, and South Africa. Trial 0021 was a double-
`blind, double-dummy, randomized, multicenter, par-
`allel-group trial conducted in North America. Recruit-
`ment for both trials occurred between May 1997 and
`September 1999. The full methodology for each trial
`has been reported previously.13’14
`
`Patients
`
`All patients were postmenopausal women with locally
`advanced or metastatic breast carcinoma that pro-
`gressed after adjuvant endocrine therapy [primarily
`with tamoxifen) or after first-line endocrine therapy
`for advanced disease. All women had tumors with
`
`evidence of hormone sensitivity (i.e., 2 12 months of
`adjuvant hormonal therapy before recurrence or tu-
`mor remission or stabilizafion from hormonal therapy
`for at least 3 months before progression ir1 patients
`with advanced disease or known ER or PgR positivity);
`a life expectancy > 3 months; and, in the opinion of
`the investigator, were deemed appropriate candidates
`for subsequent hormonal therapy.
`The main inclusion criteria were as follows: a
`
`World Health Organization performance status 5 2,
`histologic or cytologic confirmation of breast carci-
`noma with objective evidence of recurrence or pro-
`gression of disease, and the presence of at least 1
`measurable or evaluable (nonmeasurable) lesion. All
`
`Astrazeneca Ex. 2061 p. 2
`
`

`
`patients had to be postmenopausal (i.e., age 60 years
`or older, or 45 years or older age with arnenorrhea for
`> 12 months or follicle-stimulating hormone levels
`within postmenopausal range, or previous bilateral
`oophorectomy). Patients were excluded if they had
`received prior treatment for breast carcinoma with
`fulvestrant or an Al or if they had received prior ex-
`tensive endocrine treatment (more than one prior en-
`docrine treatment) for ABC. Other factors that resulted
`in exclusion included extensive radiation therapy
`within the previous 4 weeks (2 30% of bone marrow;
`e.g., the whole pelvis or half of the spine) or cytotoxic
`treatment within the past 4 weeks, estrogen replace-
`ment therapy within 4 weeks of randomization, treat-
`ment with luteinizing hormone-releasing hormone
`analogs within the 3 months before randomization, or
`any concurrent medical illness or laboratory abnor-
`malities that would compromise safety or prevent in-
`terpretation of results.
`Bisphosphonate treatment was permitted and, al-
`though initiation of treatment during the trial was
`discouraged, was allowed in the absence of disease
`progression. Bone lesions in patients who received
`bisphosphonates that were initiated before or after
`trial entry were evaluable for progression only.
`Patients in both trials were withdrawn from trial
`
`treatment at the discretion of the investigator if they
`had an unacceptable adverse event (AE); if noncom-
`pliance with the protocol was demonstrated; or if the
`patient was unwilling or unable to continue in the trial
`or had clinical findings (including disease progres-
`sion) that conflicted with the trial protocol. All pa-
`tients were monitored for progression and survival
`after they withdrew (unless consent was withdrawn).
`All patients gave written informed consent, and the
`relevant ethics committees approved the studies.
`
`Trial Design
`Patients were randomized to receive either fulvestrant
`
`250 mg (1 X 5 mL on Trial 0020 or 2 X 2.5 mL on Trial
`0021; n = 428) i.m. once monthly or anastrozole 1 mg
`(n = 423) orally once daily. Patients received the treat-
`ment to which they were randomized until there was
`objective evidence of disease progression or until
`withdrawal from the trial. The trial treatment was then
`
`stopped, standard therapy was initiated, and the pa-
`tients were monitored until death.
`
`The primary endpoint was time to progression
`(TTP). Secondary endpoints included the objective re-
`sponse (OR) rate (defined as complete response [CR]
`+ partial response [PR] using the Union Internation-
`ale Contre le Cancer criteria),15 duration of response
`(DOR), time to treatment failure (TTF), time to death
`(TTD), and tolerability. Clinical benefit (CB: CR + PR
`
`Fulvestrant VS. Anastrozole in Breast CA/Robertson et al.
`
`231
`
`+ stable disease [SD] 2 24 weeks) and duration of CB
`also were determined.
`
`Trial Treatments
`
`Fulvestrant was supplied as a single-dose, oily, 5%
`solution; and anastrozole was supplied as round,
`white, film-coated tablets. In Trial 0020, treatrnent was
`open label, and fulvestrant 250 mg was administered
`as a single, 5-mL injection into the buttocks. Because
`Trial 0021 was double blind, patients who received
`fulvestrant also received daily oral placebo tablets,
`and patients who received anastrozole also received
`monthly placebo i.m.
`injections. In Trial 0021,
`the
`fulvestrant dose or placebo was given as 2 2.5 mL
`injections, with 1 injection into each buttock.
`
`Statistical Methods
`
`The trials were designed to detect the superiority of
`fulvestrant 250 mg in terms of efficacy and tolerability
`compared with anastrozole 1 mg in postmenopausal
`women with ABC. For each trial, the final analysis was
`scheduled to occur when 340 events (i.e., objective
`disease progression or death) had occurred across the
`2 groups. This would provide 90% power to detect a
`hazard ratio (HR) 2 1.43 or S 0.70 for fulvestrant
`treatment compared with anastrozole treatment, at a
`significance level of 5%. To achieve the required num-
`ber of events, the plan was to recruit 392 pafients (196
`patients in each treatment group) into each of the 2
`trials. In addition to the separate analysis of each
`trial,13’14 a prospective plan to undertake a combined
`analysis to provide more precise estimates of the treat-
`ment effects was made.
`
`Data on the efficacy parameters were analyzed
`and summarized on an intem‘z'0n—t0—treat basis. The
`
`protocols for these trials originally contained a fulves-
`trant 125 mg treatment arm. Because this dose had
`not been tested clinically, a preliminary summary was
`performed when 30 patients (across both trials) had
`been treated for 3 months with fulvestrant 125 mg. At
`that time, the lack of an OR in any patient resulted in
`dropping the treatment arm from the study. In addi-
`tion, an interim analysis of TTP and OR was per-
`formed when 170 events had occurred in each trial.
`
`Because of this interim analysis, statistical significance
`levels for TTP and OR were adjusted from 5.0% to
`4.86% (and confidence limits were adjusted from 95%
`to 95.14%). All significance levels are two-sided.
`
`Time to progression
`TTP was defined as the number of days from the date
`of randomization until the date of objective disease
`progression or until death from any cause, which ever
`occurred first. Death was regarded as a progression
`
`Astrazeneca Ex. 2061 p. 3
`
`

`
`232
`
`CANCER July 15, 2003 / Volume 98 I Number 2
`
`event in patients who died prior to disease progres-
`sion. For patients did not have disease progression at
`the time of data cut-off, data were right censored to
`the date of the last assessment to allow analysis.
`Treatments were compared using a Cox propor-
`tional hazards regression model and included the fol-
`lowing covariates: trial, age, performance status, mea-
`surable
`disease
`compared with nonmeasurable
`disease, receptor status, previous response to hor-
`mone therapy, previous use of cytotoxic chemotl1er-
`apy, and use of bisphosphonate therapy for bone dis-
`ease. A global test using a 1% significance level was
`performed to determine whether there were signifi-
`cant treatment-by-baseline covariate interactions by
`considering a model that contained all treat1nent-by-
`baseline covariate interactions apart from trial. In ad-
`dition, a separate test for the presence of an interac-
`tion between trial and treatment also was undertaken
`
`using a 5% significance level. Both tests were nonsig-
`nificant. A nonsignificant treatment-by-trial interac-
`tion test indicated that it was appropriate to combine
`the trials. Estimates of the treatment effects are ex-
`
`pressed as HRs together with the corresponding con-
`fidence intervals (Cls) and P values. TTP also was
`summarized using Kaplan—Meier curves for each
`treatment group, and the median TTP was calculated.
`
`Best objective response
`Each patient was assessed for their OR at each visit to
`the clinic. A best OR of CR was assigned if a patient
`had no clinical, radiologic, or biochemical evidence of
`residual lesions on 1 visit with no evidence of disease
`
`recurrence or death within the subsequent 4 weeks. A
`best OR of PR was assigned when disease progression
`was not evident and disease was improved compared
`with the baseline assessment, with no evidence of
`disease recurrence or death within the subsequent 4
`weeks.
`
`The proportions of patients who had an OR
`were compared across the two treatments using a
`logistic regression model (with the same covariates
`that were used for TTP). A global test using a 1%
`significance level was performed to determine
`whether there were significant treatment-by-base-
`line covariate interactions by considering a model
`that contained all of the treatment-by-baseline co-
`variate interactions apart from trial. In addition, a
`separate test for the presence of an interaction be-
`tween trial and treatment also was undertaken using
`a 5% significance level. Both tests were nonsignifi-
`cant, and the nonsignificant treatment-by-trial in-
`teraction test indicated that it was appropriate to
`combine the trials.
`
`Fulvestrant was compared retrospectively with
`
`anastrozole for noninferiority with respect to OR and
`TTP using a one-sided CI of 95.57%. These limits were
`identical to using the upper limit of the 95.14%, two-
`sided CI for the analysis of TTP and the lower limit of
`the 95.14%, two-sided CI for the difference in response
`rates of OR. Based on the historic performance of
`anastrozole (compared with megestrol acetate) of a
`median TTP of approximately 5 months, the criterion
`for noninferiority was established by an independent
`group of experts who agreed that the two-sided 95% CI
`for the TTP HR should allow a median TTP of < 4
`
`months for inferiority of fulvestrant to anastrozole.
`The requirement for showing noninferiority for TTP,
`therefore, was based on an upper one-sided confi-
`dence limit for the TTP HR not greater than 1.25, thus
`ruling out a deficiency of 25% for the experimental
`treatment. This criterion was used previously for
`United States regulatory submissions of hormonal
`treatments for patients with ABC. In the saine submis-
`sions, the requirement for demonstrating noninferior-
`ity in terms of response rate was based on ruling out a
`deficiency in the difference in response rates of > 10%
`(upper one-sided CI not greater than 1.10). Conse-
`quently, these criteria were used to assess the nonin-
`feriority of fulvestrant relative to anastrozole in the
`current trial.
`
`Time to treatment failure
`
`TTF was defined as the number of days from ran-
`domization until the earliest occurrence of disease
`
`progression, death from any cause, or withdrawal
`from treatment. For assessment purposes, data from
`patients who did not have treatment failure at the
`time of data cut-off were right censored to the last
`assessment date. Any patient who did not receive
`any trial therapy was assigned an uncensored TTF of
`0 days. Statistically, TTF was analyzed using a
`method similar to that used to analyze TTP. The
`tests for treatment-by-covariate interactions were
`not significant, and the nonsignificant treatment-
`by-trial interaction test indicated that it was appro-
`priate to combine the trials.
`
`Duration of response
`The median DOR at 22.1 months of follow-up was
`calculated only for patients who had an OR. DOR was
`defined as the number of days from randomization
`until the first day on which disease progression was
`observed. Patients who died before they reached pro-
`gression were classed as completing their response at
`time of death. The DOR was summarized using
`Kaplan—Meier curves for each treatment group, and
`the median DOR also was calculated for each group.
`In addition, a statistical analysis of DOR was per-
`
`Astrazeneca Ex. 2061 p. 4
`
`

`
`formed using all randomized patients (defined for re-
`sponders as the time from onset of response to disease
`progression and, for nonresponders, as zero).
`
`Duration of clinical benefit
`CB was defined as the achievement of an OR or of SD
`
`> 24 weeks. For patients who achieved CB, the dura-
`tion of that benefit was calculated as the time between
`the date of randomization and the first date when
`
`disease progression was observed or when death oc-
`curred. Data for CB were summarized in the same
`manner as data for DOR.
`
`Time to death
`
`The protocol called for analyzing the TTD when > 50%
`of pafients had died. At the time of data analysis, only
`35.6% of patients had died: Therefore, no formal sta-
`tistical analyses were conducted for TTD.
`
`Tolerability
`All safety data were listed and summarized according
`to the treatment received. AEs were presented using
`the Coding Symbols for Thesaurus ofAdverse Reaction
`Terms terminology. At the outset of the trial, seven AEs
`that were considered relevant to endocrine therapy
`were predefined for statistical analysis. These were
`gastrointestinal disturbances, hot flashes, vaginitis,
`weight gain, thromboembolic disease, urinary tract
`infection, and joint disorders (including arthralgia, ar-
`throsis, and arthritis). The analysis of the predefined
`AEs was performed using a logistic regression analysis.
`Results are presented as an ORs, 95% confidence lim-
`its, and Pvalues.
`
`RESULTS
`Patient Characteristics
`
`The intention-to-treat population for the current
`combined analysis was 851 patients, including 428
`patients in the fulvestrant 250 mg group and 423
`patients in the anastrozole 1 mg group. The majority
`of patients (96% in the fulvestrant group and 97% in
`the anastrozole group) had been treated previously
`with tamoxifen, and a few had received megestrol
`acetate (0.70% in the fulvestrant group and 0.71% in
`the anastrozole group) and droloxifene (0.93% in the
`fulvestrant group only).
`Characteristics of the patients in the two treat-
`ment groups are shown in Table 1. The fulvestrant-
`treated and anastrozole-treated groups were matched
`well in terms of age, weight, breast carcinoma history,
`prior therapy, extent of recurrent disease, and ER/PgR
`status. Patients in Trial 0021 were slightly heavier (ful-
`vestrant group: mean weight, 71.2 kg; anastrozole
`group: mean weight, 72.7 kg) compared with patients
`
`Fulvestrant VS. Anastrozole in Breast CI-VRODOHISOH et al.
`
`233
`
`TABLE 1
`
`Demographic Characteristics of Patientsa
`
`Combined studies
`(Trials 0020 and 0021)
`
`Fulvestrant
`250 mglmonth
`(n = 428)
`
`Anastroznle
`1 mg/day
`(n = 423)
`
`Characteristic
`
`No. (98)
`
`No. ("/1)
`
`486 (Yrs)
`ean
`ange
`Weight (kg)
`ean
`ange
`Prior treatment
`Cytotoxic chemotherapy
`Endocrine therapy for advanced
`disease
`Adjuvant endocrine therapy
`Hormone receptor status
`ER and/or PgR positive
`ZR/PgR status unknown
`SR/PgR negative
`Metastatic or recurrent disease at
`baseline
`3reast
`Skin
`3one
`river
`sung
`,yrnph nodes
`Other
`Ex ent of metastatic or recurrent disease
`at baseline
`Soft tissue only
`3one only
`Visceral only
`,yrnph node only
`ot recorded
`ixedl’
`easurablelesions°
`onmeasurable lesions
`
`63 —)
`33-89 —)
`
`70 —)
`37-127 —)
`
`63 —)
`33-94 —)
`
`70 —)
`40-134 —)
`
`223 52.1)
`
`236 551
`243 568
`
`342 79.9)
`64
`5.0)
`22 5.1)
`
`29 3.8)
`83
`9.4
`205 47.9
`95 22.2
`119 27.8
`136 31.8
`49
`1.4
`
`23 5.4)
`85
`9.9
`69
`6.1
`37 8.6)
`1
`.2)
`213 19.8
`245 57.2
`183 42.8
`
`220 52.0)
`
`226 53.4
`235 556
`
`352 83.2)
`52 12.3)
`19 4.5)
`
`38 90)
`76 18.0)
`202 47.8)
`101 23.9)
`120 28.4)
`139 32.9)
`26 61)
`
`21 5.0)
`83 19.6)
`86 20.3)
`38 90)
`3 0.7)
`192 45.4)
`249 58.9)
`174 41.1)
`
`ER: estrogen receptor; PgR: progesterone receptor.
`“ Patients may have been in more than one category.
`‘’ Mixed was defined as breast and /or a combination of skin, bone, liver, lung or lymph nodes.
`° Measurable lesions were lesions that were measurable clinically in 2 perpendicular axes with at least
`1 dimension that measured :25 cm or measurable using imaging in 2 perpendicular axes with at least
`1 dimension that measured 21.0 cm.
`
`in Trial 0020 (fulvestrant group: mean weight, 68.9 kg;
`anastrozole group: mean weight, 67.8 kg). Prior use of
`cytotoxic chemotherapy was more common among
`patients in Trial 0021 than among patients in Trial
`0020 (63% vs. w 43%, respectively), and more patients
`in Trial 0020 had unknown ER and PgR status (Table
`2).
`
`Astrazeneca Ex. 2061 p. 5
`
`

`
`234
`
`CANCER July 15, 2003 / Volume 98 I Number 2
`
`TABLE 2
`
`Contrasting Levels of Unknown Hormone Receptor Status between Patient Groups in the Two Trials
`Trial 0021
`Trial 0020
`
`Anastrozole
`Fulvestrant
`Anastrozole
`Fulvestrant
`1 mg (n = 229)
`250 mg (n = 222)
`1 mg (n = 194)
`250 mg (n = 206)
`Unknown hormone T T T T
`receptor status
`No. (%)
`No. (%)
`No. (96)
`No. (%)
`
`ER unknown
`PgR unknown
`ER and PgR unknown
`
`l3 6.3
`18 (8.7)
`13 (6.3)
`
`ER: estrogen receptor; Pglt: progesterone receptor.
`
`l6 (8.2)
`26 (
`15
`
`51 (23.0)
`88 (39.6)
`51 (23.0)
`
`37 (16.2)
`74 (32.3)
`37 (16.2)
`
`TABLE 3
`
`1.0
`
`Progression Status throughout flie Period of Treatment with Once-
`Monthly Intramuscular Fulvestrant (250 mg) or Once-Daily Oral
`Anastrozole (1 mg)
`
`Fulvestrant
`(n = 428)
`
`Anastrozole
`(n = 423)
`
`Progression status
`
`N0 disease progression
`Total with disease progression
`Progression during treatment
`Progression after treatment withdrawal
`Death before progression
`
`No. (91)
`
`73 (17.1
`355 (82.9)
`332 (77.6
`5 (1.2)
`18 (4.2)
`
`No. (%)
`
`65 (15.4)
`358 (84.6)
`335 (79.2)
`8 (1.9)
`15 (3.5)
`
`Progression Status
`Patients were followed for a median of 15.1 months
`
`from the date of randomization. Progression status
`throughout the period of treatment is shown in Table
`3. At the time of analysis, 355 patients (82.9%) in the
`fulvestrant group and 358 patients (84.6%) in the anas-
`trozole group had disease progression, with 17.1% and
`15.4% of patients, respectively, remaining progression
`free: The difference between treatment groups was
`not statistically significant. Death occurred before tu-
`mor progression in 18 patients (4.2%) in the fulves-
`trant group and in 15 patients (3.5%) in the anastro-
`zole group.
`
` Proportion
`
`withoutprogression
`
`— Fulvestrant 250 mg
`- - Anastrozole 1 mg
`
`600
`1000
`1200
`400
`E00
`Time to progression (days)
`
`FIGURE 1. Kaplan—Meier curve of the probability of time to progression
`(ITP). Estimated TTP for patients receiving fulvestrant was 5.5 months, com-
`pared with 4.1 months for patients receiving anastrozole (hazard ratio, 0.95;
`95% confidence interval, 0.82-1.10; P = 0.48).
`
`TABLE 4
`Median Time to Disease Progression for Phase III Trials 0020 and
`0021 Individually and in a Prospective Combined Analysis
`
`TTP (mos)
`
`Analysis
`
`Fulvestrant
`
`Anastrozole
`
`Trial 0020
`Trial 0021
`Combined analysis
`
`5.5
`5.4
`5.5
`
`5.1
`3.4
`4.1
`
`HR
`
`0.98
`0.92
`0.95
`
`CI
`
`P value
`
`0.80-1.21
`0.74-1.14
`0.82-1.10
`
`0.84
`0.43
`0.48
`
`TTP: time to progression; HR: hazard ratio; Cl: confidence interval.
`
`Time to Progression
`The estimated median TTP was 5.5 months in the
`
`fulvestrant group, compared with 4.1 months in the
`anastrozole group (HR, 0.95; 95.14% CI, 0.82-1.10; P
`= 0.48). These data demonstrate noninferiority of ful-
`vestrant relative to anastrozole for TTP. Kaplan—Meier
`plots of the overall TTP data are shown in Figure 1.
`The median TTP values for Trial 0020 and Trial 0021
`
`together with the median TTP values
`individually,
`from the combined analysis, are shown in Table 4.
`
`Time to Treatment Failure
`The estimated median TTF was 4.6 months for fulves-
`
`trant and 3.6 months for anastrozole, although this
`difference was not statistically significant (HR, 0.96;
`95% CI, 0.83-1.11; P = 0.61). The majority of treatment
`failures were due to objective disease progression (ful-
`vestrant, 342 patients [93.4% of patients who failed
`treatment]; anastrozole, 350 patients [95.6% of pa-
`tients who failed treatment]). Other reasons for treat-
`
`Astrazeneca Ex
`
`.2061 p. 6
`
`

`
`TABLE 5
`Best Objective Responses to Once-Monthly Intramuscular Fulvestrant
`(250 mg) and Once-Daily Oral Anastrozole (1 mg)
`
`Fulvestrant
`(n = 428)
`
`Anastrozole
`(n = 23)
`
`Fulvestrant VS. Anastrozole ll] Breast CA/RODOHISOH et al.
`
`235
`
`o N
`
`
`
`Proportionresponding
`
`— Fulvesirant 250 mg
`-— Anastrozole 1 mg
`
`
`0
`6
`12
`18
`24
`Duration of response (mos)
`
`30
`
`36
`
`40
`
`FIGURE 3. Kaplan—Meier estimates for duration of response (DOR) from the
`onset of response to disease progression (all patients). Ratio of average DORs
`(fulvestrantzanastrozole) = 1.30 (95% confidence interval, 1.13-1.50; P
`< 0.01).
`
`1.0
`0 Q
`0 8
`0.7
`0.6
`0.5
`0.4
`0.3
`0.2
`0.1
`
`
`
`Proportionresponding
`
`
`
`— Fulvestrant 250 mg
`-— Anastrozolel mg
`
`Duration of clinical benefit (mos)
`
`40
`
`FIGURE 4. Kaplan—Meier estimates for duration of clinical benefit (DOCB).
`Median DOCB for patients receiving fulvestrant was 11.8 months, compared
`with 11.2 months for patients receiving anastrozole.
`
`No. (%)
`
`11 (2.6)
`59 (13.9)
`70 (16.5)
`103 (24.3)
`4 (0.9)
`25 (5.9)
`221 (52.2)
`353 (83.5)
`
`173 (40.9)
`
`Response
`
`CR
`PR
`Total CR and PR
`SD 2 24 wks
`SD < 24 wks
`N0 disease progression
`Disease progression
`Total
`Clinical benefit (CR + PR +
`SD 2 24 wks)
`
`No. (%)
`
`20 (4.7)
`62 (14.5)
`82 (19.2)
`104 (24.3)
`6 (1.4)
`21 (4.9)
`215 (50.2)
`346 (80.8)
`
`186 (43.5)
`
`CR: complete response; PR: partial response; SD: stable disease.
`
`..
`
`'.
`
`1.0
`0.9
`0.8
`0.7
`0.0
`0.5
`0.4
`0.3
`
`
`
`Proportionresponding
`
`1.
`
`" -._
`
`'
`
`‘-- -.
`
`.
`
`-- Anasirozole1 mg 0
`
`— Fulvestrant 250 mg
`
`6
`
`12
`
`24
`18
`Duration of response (mos)
`
`30
`
`36
`
`FIGURE 2. Kaplan—Meier estimates for duration of response (DOR) from
`randomization to disease progression in responding patients. Median DOR for
`patients receiving fulvestrant was 16.7 months, compared with 13.7 months
`for patients receiving anastrozole.
`
`ment failure included AEs (fulvestrant, 6 patients
`[1.4%]; anastrozole, 5 patients [1.2%]), protocol non-
`compliance (fulvestrant, 6 patients [1.4%]; anastro-
`zole, 5 patients [1.2%]), and withdrawal of informed
`consent (fulvestrant, 5 patients [l.2%]; anastrozole, 2
`patients [0.5%]).
`
`Objective Response Rate
`The best ORS for patients in the fulvestrant and anas-
`trozole groups are shown in Table 5. The OR rate was
`19.2% for the fulvestrant group, compared with 16.5%
`for the anastrozole group. The difference in response
`rates was 2.75% (95.l4% CI, 2.27—9.05%; P = 0.31). The
`lower bound of the CI for the difference in response
`rates was greater than -10% and therefore satisfied
`the criteria for the noninferiority of fulvestrant relative
`to anastrozole.
`
`Duration of Response
`Extended follow-up (median, 22.1 months) was per-
`formed to obtain more complete information for DOR.
`The median DOR, as measured from randomization to
`progression, in patients who responded to treatment
`was 16.7 months for the fulvestrant group (n = 84) and
`13.7 months for the anastrozole group (n = 73) (Fig. 2).
`In the statistical analysis of DOR, which included all
`randomized patients, with DOR defined from the on-
`set of response to disease progression for responders
`and as 0 for nonresponders, the DOR was significantly
`longer for patients in the fulvestrant group compared
`with patients in the anastrozole group. The ratio of
`average response durations was 1.30 (95% CI, 1.13-
`1.50; P < 0.01). Kaplan—Meier curves for DOR in all
`randomized patients are shown in Figure 3.
`
`clinical Benefit Rates
`The CB rates achieved with fulvestrant and anastro-
`
`zole are shown in Table 5. CB rates (CR + PR + SD
`2 24 weeks) were 43.5% for the fulvestrant group (n
`
`Astrazeneca Ex. 2061 p. 7
`
`

`
`236
`
`CANCER July 15, 2003 / Volume 98 I Number 2
`
`TABLE 6
`Proportion of Patients with Predefined Adverse Events
`Fulvestrant
`Anastrozole
`(n = 423)
`(n = 423)
`
`Event
`
`No. (96)
`
`No. (%)
`
`P value
`
`Gastrointestinal disturbances“
`Hot flashes
`Joint disorders
`Thromboembolic disease
`Urinary tract infection
`Vaginitis
`Weight gain
`
`196 (46.3)
`89 (21.0)
`23 (5.4)
`15 (3.5)
`31 (7.3)
`11 (2.6)
`4 (0.9)
`
`185 (43.7)
`87 (20.6)
`45 (10.6)
`17 (4.0)
`18 (4.3)
`6 (1.9)
`7 (1.7)
`
`0.53
`0.91
`0.0036
`0.68
`0.06
`0.51
`0.35
`
`“Gastrointestinal disturbances included anorexia constipation, diarrhea nausea and emesis.
`
`= 186) and 40.9% for the anastrozole group (n = 173),
`with the analysis showing no statistically significant
`difference (difference in CB rates, 2.34%; 95% CI, —
`4.42% to 9.36%; P = 0.51). The median duration of CB
`was 11.8 months for the fulvestrant group (n = 187)
`compared with 11.2 months for the anastrozole group
`(n 2 174) (Fig. 4).
`
`Time to Death
`
`At the time of data analysis, 303 patients (35.6%) had
`died: 155 patients (36.2%) in the fulvestrant group and
`148 patients (35.0%) in the anastrozole group. Because
`insufficient numbers of patients had died to permit a
`meaningful comparison of survival data (< 50% of
`patients in eithe