Breast Cancer Research and Treatment 73: 161—175, 2002.
`1* © 2002 Kluwer Academic Publishers. Printedin the Netherlands.
`
`Report
`
`Phase III randomized trial of droloxifene and tamoxifen as first-line
`
`endocrine treatment of ER/PgR-positive advanced breast cancer
`
`A. Buzdarl, D. Hayesz, A. El— Khoudary3, S. Yan4, P. Lonnings, M. Lichinitser6, R. Gopal7,
`G. Falksong, K. Pritchardg, A. Lipton“), K. Wolter11,A. Lee11,K.F1y11,R.Chew11,
`M. Alderdice“, K. Burke”, and P. Eisenberg13 *
`1The University of Texas, MD. Anderson Cancer Center, Houston, TX; 2Georgetown University Medical Center,
`Washington, DC, USA; 3National Cancer Institute, Cairo University, Cairo, Egypt; 4Cancer Hospital of the
`Chinese Academy of Medical Science, Beijing, China; 5Haukeland Sykehus, Bergen, Norway; 6Cancer Research
`Center, Moscow, Russia; 7Tata Memorial Hospital, Bombay, India; 8 University ofPretoria and Pretoria Academic
`Hospitals, Pretoria, South Africa; 9Toronto-Sunnyhrook Regional Cancer Center, Toronto, Ontario, Canada;
`10M. S. Hershey Medical Center, Hershey, PA; 11Pfizer Central Research, Groton, CT, USA; 12Klinge Pharma
`GmbH, Munich, Germany; 13Sutter/CHS Cancer Research Group, Greenbrae, CA, USA
`
`Key words: advanced breast cancer, droloxifene, positive hormone receptors, randomized trial, tamoxifen
`
`Summary
`
`Purpose: This trial was designed to demonstrate equivalence between droloxifene 40 mg/d and tamoxifen 20 mg/d
`as first-line treatment in pre- and post-menopausal women with ER+ and/or PgR+ advanced breast cancer based
`on time to disease progression and tumor response.
`Materials and methods: One thousand three hundred fifty four women with measurable disease, previously
`untreated by hormonal or chemotherapy for advanced or recurrent breast cancer, were enrolled by 179 institutions
`in 35 countries. Patients were stratified at baseline for menopausal status. Patients receiving adjuvant hormonal
`therapy within 1 year were excluded. All patients gave written informed consent, were randomized to 40 mg
`droloxifene or 20 mg tamoxifen daily as single-agent therapy and underwent tumor assessment every 3 months. A
`central committee reviewed digitized images for all cases of tumor progression or objective response.
`Results: The hazard ratio (droloxifene/tamoxifen) for the primary endpoint, time to disease progression, was
`1.287 favoring tamoxifen (95% CL: 1.114—1.487; p < .001). The objective response rate (CR + PR) was 22.4%
`for droloxifene and 28.6% for tamoxifen (p = .02). Tamoxifen was superior to droloxifene overall, among both
`pre- and postmenopausal patients and among patients 565 years; there was no difference among women >65
`years. The hazard ratio for all-cause mortality was 0.871 (95% CL: 0.672—1.129; p = .29), favoring droloxifene
`but not statistically significant.
`Conclusions: Droloxifene was significantly less effective than tamoxifen overall and particularly among
`women under 65 years. Tamoxifen and droloxifene were both less effective in pre-menopausal women with
`receptor-positive disease compared to post-menopausal women. Further clinical development of droloxifene was
`stopped.
`
`Introduction
`
`Droloxifene is a novel selective estrogen receptor
`modulator (SERM) whose potential as a treatment
`
`*For the Droloxifene 301 Study Group.
`
`for breast cancer has been suggested by several non-
`comparative studies [1—8]. Preclinical studies [9, 10]
`
`have shown that droloxifene has a shorter serum half-
`life and a higher affinity for the estrogen receptor than
`tamoxifen, an accepted first-line treatment option for
`many women with hormonally sensitive breast cancer.
`
`AstraZeneca Exhibit 2083 p. 1
`lnnoPharma Licensing LLC v. AstraZeneca AB lPR2017-00904
`Fresenius-Kabi USA LLC v. AstraZeneca AB lPR2017-01910
`
`

`

`162
`
`A Baza’ar et al.
`
`Pre-clinically, the two agents differ most with respect
`to animal carcinogenicity, with tamoxifen shown to
`cause hepatic tumors in 98% of treated animals while
`droloxifene- and control-treated animals showed a l—
`
`it is well
`2% incidence [10]. In humans, of course,
`known that tamoxifen increases the incidence of en-
`dometrial cancer but has no effect on the incidence of
`
`hepatic tumors. Nevertheless, these pre-clinical find-
`ings suggested that droloxifene might be more useful
`than tamoxifen for longer-term breast cancer therapy,
`such as in the adjuvant setting, provided that droloxi-
`fene possessed efficacy that was at least equivalent to
`tamoxifen in patients with advanced disease. A Phase
`2 study of 369 women with advanced breast cancer
`randomized to one of three doses of droloxifene (20,
`40 or 100 mg/d) as first-line hormonal therapy gave
`encouraging results, with CR + PR rates of 30, 47 and
`44%, respectively, for the three treatment groups [6].
`Droloxifene has also been studied in patients with ad-
`vanced breast cancer who have been exposed to prior
`endocrine treatment, including some who were resist-
`ant to tamoxifen, with partial response seen in 15% of
`these patients [1 l]. The purpose of the present study
`was to demonstrate equivalence between droloxifene
`40 mg daily and tamoxifen 20 mg daily by comparing
`time to disease progression in a global study intended
`to reflect the diversity of the patient population ac-
`tually using first-line hormonal therapy for advanced
`breast cancer.
`
`Materials and methods
`
`Study design
`
`This was a prospective, randomized, active-control,
`double-blind, multi-center, parallel-group comparison
`of droloxifene and tamoxifen,
`in women with ER+
`and/or PgR+ advanced breast cancer. Patients were
`stratified by menopausal status. The primary endpoint
`of the trial was time to disease progression, defined
`as the time from randomization to the first objective
`finding demonstrating a 25% increase in the size of at
`least one tumor lesion or the appearance of any new
`tumor lesion or death due to breast cancer. Overall
`
`tumor response was a secondary endpoint, along with
`response duration and various subset analyses. Time to
`progression and tumor response were determined for
`every patient by a central endpoint evaluation commit-
`tee. Parameters used to pre-define subsets of patients
`in the prospective statistical analysis plan included
`age, menopausal status, geography, disease status at
`
`baseline, adjuvant therapy, and performance status as
`listed in Table 4. The independent Data Safety Moni-
`toring Board (DSMB) conducted periodic, planned in-
`terim analyses of the data in order to monitor the safety
`of the trial.
`
`Patient selection
`
`Eligible patients included pre- or post-menopausal
`women with biopsy-proven breast cancer with dis-
`tant metastases, locoregional recurrences not suitable
`for local therapy or inoperable primary tumors. Pa-
`tients were defined as postmenopausal if menses had
`ceased for more than 1 year and serum estrogen was
`below 30 pg/ml, or if the patient had undergone bi-
`lateral oophorectomy. Acceptable target lesions were
`measurable in at least one dimension, at least 1cm
`in size and not previously radiated. Lytic bone le-
`sions not visible on plain x-ray were excluded as target
`lesions, as were any blastic bone lesions or blastic
`portions of mixed lesions. Patients were excluded
`if they had received any prior chemo- or hormonal
`therapy (including oophorectomy) for advanced dis-
`ease, or adjuvant hormonal therapy within the past
`year or adjuvant chemotherapy within the past month
`prior to randomization. Patients were required to have
`receptor-positive tumors defined as ER+ and/or PgR+
`(unknown receptor status for both ER and PgR was
`not allowed). ECOG performance status of 60% or
`greater was required. Patients with brain, leptomenin-
`geal or extensive (> 1/3 of the liver) hepatic metastases
`were excluded, as were patients with hypercalcemia or
`significant risk for thromboembolic events.
`
`Pretreatment evaluation
`
`Prior to initiating study drug treatment, a complete
`history, physical exam and tumor assessment were
`performed, including bone scan, chest x-ray, and ab-
`dominal CT (or liver ultrasonography). Any suspicion
`of bone metastases on the bone scan required a defined
`set of eight skeletal plain films for confirmation. Rep-
`resentative tumor lesions were identified for each
`
`patient. Tumor response or progression would be de-
`termined based upon changes in these target lesions.
`Any new lesion was deemed disease progression
`regardless of changes in target lesions.
`
`Treatment plan
`
`Each patient received, by random assignment, either
`(a) 40 mg/day droloxifene (Pfizer Central Research,
`
`AstraZeneca Exhibit 2083 p. 2
`
`

`

`Phase III randomized trial of droloxifene versus tamoxifen
`
`163
`
`Groton, CT) + placebo tamoxifen or (b) 20 mg/day
`tamoxifen (Tamoxpuren®, Klinge Pharma, Munich,
`Germany) + placebo droloxifene.
`
`the trial closed, according to the investigator’s review
`of the case, were accepted as ‘no change’ in the trial
`database without committee review.
`
`Follow-up tumor assessments
`
`Ethics
`
`After randomization, patients returned to clinic every
`3 months. At each visit, physical examination, chem-
`istry/hematology, chest x-ray, and measurement of
`target lesions were performed for all patients. Ab-
`dominal CT (or hepatic sonography) and/or bone scan
`with skeletal x-ray series were performed every 3
`months for those patients with relevant target lesions
`at baseline. Patients with no target lesions in either
`bone or liver received abdominal CT and bone scan
`
`at 6 month intervals and at the end of the study.
`
`Response evaluation
`
`Target lesion measurements were recorded in a log
`every 3 months. All physical examination measure-
`ments were checked for errors between the medical
`
`record and the case report forms by monitors who vis-
`ited each study center at least every 4—8 weeks. All
`x-ray and scan images of target lesions, or new le-
`sions, were first evaluated by each investigator in order
`to determine the clinical plan for the patient. These
`films were then sent to a central imaging facility where
`each x-ray or scan was digitized for electronic review
`by an Endpoint Classification Committee (ECC) con-
`sisting of experienced investigators and radiologists
`from North America and Western Europe. Electronic
`images were viewed on a bank of four ultra-high re-
`solution monitors using software that allowed contrast
`adjustment to optimize image readability of x-rays and
`scans. The reviewers were blinded to treatment arm.
`
`Tumor response was evaluated according to WHO cri-
`teria [12], with additional requirements that (i) only
`x-ray, CT or MRI (not radionuclide bone scan) were
`used to determine response or progression in bone,
`and (ii) blastic bone lesions, or the blastic portion of
`mixed lytic/blastic bone lesions, were not considered
`for tumor response evaluation. The decision of the
`committee regarding objective tumor response and the
`date(s) related to that response was final as concerned
`the study analyses. This committee reviewed every
`case in which the investigator found CR, PR or disease
`progression, a death or a premature termination. At the
`close of the trial, all patients still receiving their as-
`signed study medication underwent a termination visit
`and complete tumor assessment. Only those active pa-
`tients whose disease status was unchanged at the time
`
`The ethical committee at each participating institution
`reviewed and approved the protocol and the informed
`consent document. Each patient gave written informed
`consent that met the requirements of FDA GCP regu-
`lations and the Declaration of Helsinki (as amended
`1975 and 1983), in addition to all local regulations in
`each country as required.
`
`Statistical methodology
`
`The statistical plan predicted that the study would need
`to enroll 1375 patients in order to observe 900 events
`(disease progressions) within 2 years. Patients were
`assigned to study treatment by a computer-generated
`randomization list after stratification by menopausal
`status. The study was designed as a non-inferiority
`trial employing the technique of repeated confidence
`intervals [22]. The trial was planned to continue until
`900 events had occurred in order to allow a deter-
`
`mination that the relative efficacy of droloxifene was at
`least 80% that of tamoxifen as measured by the hazard
`ratio for time to disease progression. The operating
`characteristics of the statistical inference were such
`
`that the power was 90% to declare the non-inferiority
`of droloxifene relative to tamoxifen if the true times to
`
`disease progression for the two drugs were not differ-
`ent and approximately 900 events had been observed.
`The statistical plan allowed stopping the trial before
`900 events for a statistically significant difference in
`efficacy, but required observation of 900 events in or-
`der to declare non-inferiority. Interim analyses were
`scheduled to occur with every 150 additional events
`and these results were provided only to the indepen-
`dent DSMB in order for them to review the progress
`and safety of the trial. However, the identity of each
`treatment arm remained coded until after the DSMB
`
`had made the decision to end the trial. The project
`medical and administrative staff, along with investi-
`gators and other study personnel, were unaware of any
`interim results.
`
`Hazard ratios are estimated from univariate propor-
`tional hazards regression (Cox) models with treatment
`as the sole predictor. Lifetime analyses are based
`on the product-limit method of Kaplan and Meier.
`Confidence intervals and p-values reported herein are
`nominal, that is, unadjusted for multiple comparisons
`
`AstraZeneca Exhibit 2083 p. 3
`
`

`

`164
`
`A Baza’ar et al.
`
`Table 1. Geographic regions (randomized patients)
`
`Africa/Mid—East
`(201 patients)
`
`Asia
`
`(337 patients)
`
`Eastern Europe
`(210 patients)
`
`Latin America
`
`(91 patients)
`
`North America
`
`(252 patients)
`
`Western Europe
`(263 patients)
`
`Egypt
`
`South Africa
`
`Israel
`
`P. R. China
`
`Hong Kong
`
`India
`
`Russia
`Hungary
`Czech
`
`Republic
`
`Mexico
`
`Brazil
`
`Canada
`
`Serbia
`Latvia
`Slovakia
`
`Argentina
`Chile
`
`United States
`
`France
`Germany
`Netherlands
`Austria
`
`United Kingdom
`Turkey
`Greece
`Spain
`
`Taiwan
`
`Belarus
`Poland
`
`Costa Rica
`Uruguay
`
`Puerto Rico
`
`Sweden
`Belgium
`Norway
`Italy
`
`or interim analyses. The log-rank test was used for
`comparisons of time to event distributions. The chi-
`square test was used for comparisons of response rates.
`All of the subgroup analyses shown in Tables 4 and
`5 were included in the prospective statistical analysis
`plan and were hypothesis-testing analyses.
`
`Results
`
`Patient characteristics
`
`A total of 1966 women with advanced breast cancer
`were screened for the trial and 1354 women were ran-
`
`domized between June, 1995 and December, 1997
`at 179 study centers in 35 countries and territories
`(Table 1).
`The pretreatment characteristics of the patients
`are shown in Table 2 according to treatment group.
`The two treatment groups showed no significant dif-
`ferences with respect to the parameters in Table 2,
`with the exceptions that the patients in the drolox-
`ifene group were somewhat more likely (p<.05;
`chi-square) to have four or more tumor lesions at
`baseline and to have received prior adjuvant hormonal
`or radiation therapy.
`The mean duration of therapy was 196 days (range:
`8—920 days) in the droloxifene group and 218 days in
`the tamoxifen group (range: 6—969 days). A total of
`
`69 patients discontinued the study before disease pro-
`gression occurred. Reasons for early termination are
`shown in Table 3. Under intention-to-treat principles,
`all randomized patients were included in the analyses
`of safety and efficacy.
`
`Central review of endpoints
`
`The ECC (ECC; A. Buzdar, Chairman) centrally re-
`viewed 1026 patients for tumor response out of 1354
`enrolled patients. A total of 328/1354 cases were re-
`ported by the investigator to be ‘no change’ at the time
`of the study data cut-off (February 1998); these cases
`were not submitted for review by the ECC. Every case
`involving disease progression or complete or partial
`response (CR or PR), as judged by the investigator,
`was reviewed centrally. The committee determined the
`nature of the response and the date of response or pro-
`gression for the purposes of the analysis. The ECC
`was unable to adjudicate 36/1026 cases (3.5%) due to
`insufficient data.
`
`Disease progression
`
`Time to disease progression was the primary end-
`point of the study. Figure 1 shows the time-course
`of disease progression for all randomized patients
`by treatment group. More than half of the patients
`(744/1354) experienced disease progression during the
`
`AstraZeneca Exhibit 2083 p. 4
`
`

`

`Phase III randomized trial of droloxiferie versas tamoxifen
`
`165
`
`onoE<
`
`aoEoE<
`
`~2qu
`
`quad
`
`Sm<
`
`.m.2\woiw<
`
`omogm
`
`“mam
`
`omohsm
`
`$03
`
`NmN
`
`20
`
`hmm
`
`20m
`
`02m
`
`m0N
`
`$2
`
`$Nw
`
`$Eu
`
`$55
`
`$0$mm
`
`$2
`
`$00
`
`$3
`
`$Nm
`
`$hm
`
`$2
`
`$mw
`
`$2
`
`$0
`
`$0w
`
`$02
`
`$2
`
`$2
`
`$mm
`
`$mm
`
`$mm
`
`$m
`
`$0
`
`$w
`
`$0m
`
`$0w
`
`$hw
`
`$2
`
`$0
`
`$5
`
`$wm
`
`$0m
`
`$0w
`
`$2
`
`$0
`
`$2
`
`$2
`
`$2
`
`$2
`
`$0
`
`$02
`
`$m0
`
`$wm
`
`$5»
`
`$wN
`
`$02
`
`$ww
`
`mm
`
`mm
`
`
`
`
`
`
`
`QBSGE023550mmoE:3:0an.08:2oncan050%E0830:NE3:0an2038:005200monmifiofiano308305on.Nm3£
`
`
`
`
`
`
`
`
`
`nowunofiflk
`
`050208282
`
`mn0
`
`00
`
`0mm
`
`Dam
`
`m2
`
`0mm
`
`N.m0
`
`mum
`
`mmw
`
`m2
`
`ww
`
`mm
`
`0mm
`
`3m
`
`
`
`:3ng“0386“nogoom
`
`EcoEnowwohoog
`
`$3388
`
`:0
`
`m2
`
`Dam
`
`02m
`
`wm2
`
`Nwm
`
`0.00
`
`hum
`
`M02
`
`mmw
`
`m2
`
`0N
`
`Du
`
`00
`
`Ewe»mw50:2
`
`28%w0lmw
`
`28%+m0
`
`3:0an00.oZ
`
`0388253
`
`ow<
`
`2330052592
`
`2339058on
`
`29%2330982
`
`AmESwEEV
`
`
`
`EwEB:82
`
`$002I0w
`
`$0>I00
`
`$00V
`
`85>?
`
`554$
`
`Mona
`
`$50
`
`32>on82
`
`
`
`808\Eflofiam
`
`832
`
`
`“0386Bounced
`2583“aowmofim
`
`AstraZeneca Exhi it 2083 p. 5
`
`

`

`166
`
`A Buzdar et al.
`
`8:0::<
`
`00:0::<
`
`:fioz
`
`:93
`
`$H
`
`$3
`
`$0N
`
`$wN
`
`$m
`
`$N
`
`$0w
`
`$0m
`
`$w
`
`$w0
`
`$0
`
`$0w
`
`$0N
`
`$mN
`
`$w:
`
`$mm
`
`$0
`
`$0:
`
`$0w
`
`$w
`
`$50
`
`$0N
`
`$w
`
`$m0
`
`$wm
`
`$H
`
`$m0
`
`$m
`
`$3
`
`$0m
`
`$3
`
`$m
`
`$w5
`
`$N
`
`$0»
`
`$w0
`
`$N
`
`$Nw
`
`$0:
`
`$N
`
`$H
`
`$mw
`
`$2
`
`$55
`
`$w
`
`$w:
`
`$H
`
`$:
`
`Sm<
`
`.m.2\00::w<
`
`000::m
`
`$5
`
`$0:
`
`$5N
`
`$5N
`
`$mm
`
`$0?
`
`$5
`
`$w
`
`$N
`
`$w5
`
`$mN
`
`$mN
`
`$5N
`
`$N0
`
`$m
`
`$5
`
`$mm
`
`$0w
`
`$05
`
`$3
`
`$0
`
`$m0
`
`$m
`
`$mm
`
`$mw
`
`$m
`
`$3
`
`$w5
`
`$m
`
`$0:
`
`$5
`
`$m
`
`$mN$m5
`
`$N
`
`$Nw
`
`$w:
`
`$0
`
`$0w
`
`$2
`
`$mm
`
`$5?
`
`$5
`
`$3
`
`$0w
`
`$H
`
`$0
`
`$w0
`
`$0
`
`$w5
`
`$NN
`
`$0
`
`“mam
`
`000::m
`
`“£03
`
`50:08:35
`
`0:000:28Q
`
`$wN
`
`$0:
`
`$N
`
`$0m
`
`$0?
`
`$H
`
`$0
`
`$0
`
`$05
`
`$wN
`
`$3
`
`$0
`
`$N5
`
`$m
`
`$0
`
`$5w
`
`$w
`
`$0w
`
`$0:
`
`$w
`
`0:
`
`mm:
`
`mm
`
`m:
`
`03
`
`NNO
`
`Hm
`
`N0?
`
`:N
`
`mm
`
`00
`
`m0w
`
`Hm
`
`m5
`
`3wwm
`
`M00»
`
`E:0:
`
`m0
`
`0w:
`
`vm
`
`wa
`
`:5
`
`0:0
`
`05
`
`5:
`
`mm
`
`mm
`
`0N
`
`0N
`
`5m?
`
`
`
`E000030::0:0m
`
`
`
`30::0:00.:0:0:00:0m
`
`0006020002
`
`:MEN
`
`5m?
`
`wNN
`
`23050:::00200002
`
`02:80
`
`
`
`:000000::0wobmm
`
`23050:::00200002
`
`02:80
`
`
`
`:000000:0:0:030005
`
`
`
`5:030::0:0m
`
`
`
`£00830:0:035
`
`30:08umA
`
`0525“82
`
`0006020
`
`2:88$10
`
`RES:000.0385
`
`
`
`
`
`0:0m0E£00830::024
`
`
`
`
`
`00:A28:00.20\mHmEinH
`
`
`
`
`
`0:830£50:wa53:50
`
`£033.80:
`
`
`
`
`
`5280:£50:wa538:0
`
`5::008:0300
`
`
`
`$055053.N0305
`
`05m
`
`00w
`
`HON
`
`
`
`
`
`0:030£00830::024
`
`00060.20002
`
`80002082
`
`OHOEHOVml:
`
`
`
`E060::0::B00.0Z
`
`A traZeneca EXhi it 2083 p. 6
`
`

`

`Phase III randomized trial of droloxiferie versas tamoxifen
`
`167
`
`onoE<
`
`aoioE<
`
`:fioZ
`
`quad
`
`Sm<
`
`.m.2\woiw<
`
`omohsm
`
`$3
`
`$wH
`
`$3
`
`$3
`
`$2
`
`$05
`
`$Nw
`
`$ww
`
`$3
`
`$mw
`
`$5»
`
`$mm
`
`$mo
`
`$wm
`
`$3
`
`$mm
`
`$wm
`
`$wo
`
`$wm
`
`$5»
`
`$mm
`
`$wm
`
`$No
`
`$3
`
`$m>
`
`$9»
`
`$om
`
`$wo
`
`$8
`
`$ww
`
`$0m
`
`$9»
`
`$ON
`
`$ow
`
`$mN
`
`$ON
`
`$2
`
`$w
`
`$m
`
`$N
`
`$2
`
`$5
`
`$>m
`
`$w
`
`$9»
`
`$w
`
`$hm
`
`$w
`
`$w
`
`$Nm6
`
`$wohwm
`
`$NmE:
`
`$wowe
`
`$90:
`
`$Hwmom
`
`$0Mmum
`
`$33
`
`$09
`
`$mmmom
`
`$HHmm
`
`.MwmHomm556
`
`“mam
`
`H83
`
`omogm:ofixofifiw
`
`onofixofioufl
`
`
`
`$35553.NNEGR
`
`on
`
`hwm
`
`mm:
`
`3:
`
`wow
`
`m5»
`
`wwm
`
`Nb
`
`mum
`
`9»
`
`\mEOHooHoEBREQQBde
`
`9:5$988iowSmOZ
`
`
`
`:Bo§::\bow5m550
`
`Enofiwom
`
`
`
`\mmfiofiofiono32>onOZ
`
`\mmfiofiofiono
`
`352:0:ESE?oz
`
`Nashua“
`
`
`
`382:0:Bax/sacsH05
`
`
`
`3585:HES/5:34
`
`Nashua“
`
`a»3me05
`
`
`EwifiéHOE
`
`ixafiofiofionoEw>5€<
`
`Amos»?:8\wEQoBme
`
`NCNEEQno\Costm
`
`583
`
`8%
`
`oz
`
`
`
`8539232>on
`
`
`
`“H9988$5005H8“96609on95592magbowfiwomintEHondascam?
`
`
`
`
`
`
`
`.owmofiw“SEES5:»3:2ng@5234?
`
`AstraZeneca Exhi it 2083 p. 7
`
`

`

`168
`
`A Buzdar et al.
`
`Table 3. Reasons for early termination
`
`No. (%) of Patients
`
`Droloxifene
`
`Tamoxifen
`
`Randomized
`
`Discontinuations
`Protocol violation
`
`Lost to follow—up
`Withdrew consent
`Adverse event
`Other
`
`681
`
`37 (5.4%)
`3
`
`673
`
`32 (4.8%)
`1
`
`10
`9
`9
`6
`
`5
`8
`14
`4
`
`course of the study. The hazard ratio for disease pro-
`gression (droloxifene:
`tamoxifen) was 1.287 (95%
`CL: 1.114—1.487; p < .001), indicating a significantly
`lower risk for disease progression in the tamoxifen
`group. The median time to progression, for all pa-
`tients, was 228 days for tamoxifen and 183 days for
`droloxifene. Early disease progression (defined as pro-
`gression 53 months from randomization) occurred in
`234/613 (38%) droloxifene patients who had at least 3
`months of follow-up and in 186/623 (30%) of tamox-
`ifen patients with similar follow-up. The incidence of
`disease progression and the hazard ratios for the pre-
`defined subpopulations are shown in Table 4. Tamox-
`ifen showed a lower risk for disease progression than
`droloxifene among both pre- and post-menopausal pa-
`tients. The hazard ratios for these two strata were
`
`not statistically different, although the superiority of
`tamoxifen was somewhat more pronounced among
`the pre-menopausal stratum. One patient could not be
`classified with respect to menopausal status and was
`censored from that subgroup analysis. A multivariate
`analysis of the predefined prognostic factors showed
`that the statistical superiority of tamoxifen was still
`evident after adjustment of these factors.
`
`Survival
`
`A total of 230/1354 patients died while on the study.
`The endpoint committee reviewed each death. It was
`determined that 10/230 deaths were due to causes
`
`other than the disease under study, however, all deaths
`were included in the analysis of overall survival. Sur-
`vival is depicted by treatment group in Figure 2 for
`all patients. The hazard ratio for all-cause mortal-
`ity was 0.871, favoring droloxifene but not statistic-
`ally significant (95% C.l.: 0.672—1.129). Differences
`in survival favoring droloxifene were found in three
`subgroup analyses: patients with a disease-free in-
`
`terval >24 months (hazard ratio = 0.528, 95% Cl:
`0317—087 9); patients who received previous adjuvant
`hormonal therapy (hazard ratio = 0.424, 95% Cl:
`0193—0932); and patients who received previous ad-
`juvant chemotherapy (hazard ratio = 0.532, 95 % C.l.:
`0319—0887). These results are unadjusted for mul-
`tiple comparisons and are considered hypothesis gen-
`erating only. No subgroup analysis of survival showed
`a significant finding in favor of tamoxifen.
`
`Objective tumor response rate and duration
`
`121/681 (18%) droloxifene patients and
`Overall,
`155/673 (23%) tamoxifen patients had a confirmed
`response to treatment (PR or CR). However, these de-
`nominators include a number of patients who were
`randomized less than 6 months before the study was
`concluded and had not yet experienced a prespecified
`endpoint (CR, PR or disease progression). Therefore,
`best response data are presented for the 541 drolox-
`ifene patients and 542 tamoxifen patients who (a) had
`a documented PR or CR or (b) had a documented dis-
`ease progression or (c) in the absence of CR, PR or
`disease progression, had been observed for at least 6
`months. The best responses for this group of 1083 pa-
`tients are shown in Table 5 as a function of treatment
`
`and disease characteristics. The tumor response rates
`for both droloxifene and tamoxifen were markedly
`lower among pre-menopausal patients (5.2 and 15.5%,
`respectively) as compared to post-menopausal patients
`(27.1 and 31.7%, respectively). The remaining 271
`patients had enrolled within the final 6 months of the
`study. These patients all had stable disease with less
`than 6 months of observation and were censored from
`
`this analysis; because the study ended, no additional
`follow-up of these patients was available.
`Among the 22/261 pre-menopausal and 254/1039
`post-menopausal patients who did show a response
`(CR or PR) during the study,
`the overall hazard
`ratio (droloxifeneztamoxifen) for time to response was
`1.053 (95% Cl: 0829—1337; p =67) suggesting no
`difference between the treatments. The median time to
`
`response was approximately 90 days in each treatment
`group and was strongly influenced by the protocols
`follow-up visit schedule (visits every 3 months). How-
`ever, only 25% of patients who would eventually
`respond did so after 100 days, and less than 10% after
`180 days. The median duration of response was 452
`days for the tamoxifen group and 555 days for the
`droloxifene group (NS).
`
`AstraZeneca Exhibit 2083 p. 8
`
`

`

`Phase III randomized trial of droloxifene versus tamoxifen
`
`169
`
`
`
`Treatment group
`
`— Droloxifene
`— - Tamoxifen
`
`
`
`
`
`
`
`f
`
`l
`
`I
`
`T
`
`I
`
`I
`
`I
`
`l Y‘j
`
`l
`
`l
`
`I
`
`l
`
`I
`
`I
`
`l
`
`I
`
`l
`
`l
`
`I
`
`l
`
`I
`
`I
`
`‘l
`
`l
`
`V
`
`l
`
`v
`
`t
`
`i
`
`l
`
`I
`
`I
`
`I
`
`I
`
`I
`
`V
`
`l
`
`I
`
`l
`
`l
`
`l
`
`W
`
`T
`
`1
`
`I
`
`I
`
`r—T
`
` '
`
`0.0
`
`Proportion
`not
`
`progressed
`
`0.6 1
`
`0.4 1
`
`0.3 <
`
`0.2 l
`0.1 ]
`
`0
`
`100
`
`200
`
`300
`
`400
`
`500
`
`600
`
`700
`
`800
`
`900
`
`1000
`
`Days
`Figure I. Kaplan—Meier curve — proportion not progressed by treatment group (product limit estimator).
`
`Safety and tolerability
`
`Approximately 70% of each treatment group reported
`one or more treatment-emergent adverse events dur-
`ing the trial. The incidence rates for these events are
`shown in Table 6. Serious adverse events, defined as
`hospitalization, cancer, other life-threatening condi-
`tions and any death by current FDA Good Clinical
`Practice regulations, occurred in 110 patients in each
`group and included hypercalcemia, thromboembolic
`events and death due to breast cancer.
`
`Decision to stop trial
`
`Two interim analyses were conducted. A provisional
`decision to end the study was made by the DSMB
`after the second interim analysis (307 events) based
`on conditional power calculations that suggested that
`the test drug was likely to be declared inferior to the
`control agent. In particular,
`the superior efficacy of
`the control in the younger, pre-menopausal population
`was the underlying reason that the DSMB decided to
`stop the study early. Further analyses were conducted
`in order to confirm that no subgroup or anomaly could
`account for the emerging statistical superiority of the
`standard treatment. The trends seen at the second in-
`
`terim analysis were born out in the final results. As
`a result of these findings the trial was stopped after
`
`744 events had taken place among 1354 randomized
`patients. It was not necessary to proceed to 900 events,
`as originally planned, in order to draw a conclusion.
`
`Discussion
`
`The results of this study show that droloxifene has
`activity in hormonally sensitive advanced breast can-
`cer when given as first-line therapy, but based on the
`time to disease progression and tumor response re-
`sults, its overall efficacy is inferior to that of tamox-
`ifen. The tumor response rate for droloxifene was
`clearly less than that for tamoxifen among women un-
`der 65 years (15% vs 23%, respectively), whereas
`the two drugs showed the same efficacy (38%) among
`patients 65 or over.
`The reason both drugs were significantly less ef-
`fective among younger or pre-menopausal patients,
`despite any baseline differences, cannot be defini-
`tively determined from the results of this study.
`Overexpression of the HER-2/c-neu/c-erbB-2 proto-
`oncogene has been shown to predict lower response
`to hormone therapy and is more common among pre-
`menopausal women [19, 20]. While erbB-2 expression
`was not determined in this trial, a separate study
`of droloxifcne in 94 advanced breast cancer patients
`showed that response rate was 9% among women with
`high pretreatment circulating levels of the extracellular
`
`AstraZeneca Exhibit 2083 p. 9
`
`

`

`170
`
`A Buzdar et al.
`
`Table 4. Time to disease progression: hazard ratios (droloxifene:tamoxifen) (95% C.l.)*
`
`Cohort
`
`All Patients
`
`Age 65 or more
`Age 45—64
`Age 44 or less
`Premenopausal
`Postmenopausal
`
`Western Europe
`Eastern Europe
`Africa/Middle East
`
`Asia
`Latin America
`North America
`
`Distant metastases present
`Distant metastases absent
`
`Primary breast cancer
`Recurrent breast cancer
`
`Bone metastases only
`Not bone metastases only
`Liver metastases present
`No liver metastases
`Four or more tumor lesions
`Three or fewer tumor lesions
`
`Disease—free interval 524 mos.
`Disease—free interval >24 mos.
`
`Prev. adjuv. hormonalM
`No prev. adjuv. hormonalM
`Prev. adjuv. chemoH
`No prev. adjuv. chemoM
`Karnofsky 80—100%
`Karnofsky 60—70%
`
`N0. disease progressions/
`No. randomized (%PD)
`
`Hazard ratio (95% CI.)
`
`744/1354 (54.9)
`197/466 (42.3)
`393/677 (58.1)
`154/211 (73.0)
`173/261 (66.3)
`570/1092 (52.2)
`
`117/263 (44.5)
`100/210 (47.6)
`112/201 (55.7)
`217/337 (64.4)
`46/91 (50.5)
`152/252 (60.3)
`
`539/939 (57.4)
`167/355 (47.0)
`336/640 (52.5)
`408/714 (57.1)
`90/167 (53.9)
`633/1138 (55.6)
`111/160 (69.4)
`620/1160 (53.4)
`258/362 (71.3)
`473/958 (49.4)
`
`494/897 (55.1)
`239/437 (54.7)
`86/140 (61.4)
`322/574 (56.1)
`220/345 (63.8)
`188/369 (50.9)
`611/1150 (53.1)
`131/202 (64.9)
`
`1.287 (1.114—1.487)
`1.045 (0.790—1.383)
`1.391 (1.140—1.698)
`1.339 (0972—1845)
`1.476 (1091—1997)
`1.224 (1.039—1.443)
`
`1.350 (0937—1944)
`1.307 (0881—1940)
`1.019 (0.703—1.477)
`1.331 (1.019—1.738)
`1.870 (1036—3374)
`1.141 (1026—1949)
`
`1.337 (1128—1584)
`1.201 (0.884—1.630)
`1.297 (1.046—1.607)
`1.284 (1.057—1.560)
`1.183 (0.781—1.791)
`1.308 (1119—1530)
`1.132 (0.775—1.655)
`1.312 (1.121—1.536)
`1.305 (1.018—1.673)
`1.210 (1010—1450)
`
`1.378 (1.086—1.759)
`1.120 (0869—1444)
`1.182 (0763—1832)
`1.277 (1026—1589)
`1.558 (1192—2037)
`1.045 (0785—1392)
`1.270 (1.083—1.489)
`1.394 (0987—1968)
`
`*In some cases the total number of patients shown in complementary categories may be less than the total
`number of randomized patients (N = 1354) due to unclassified patients.
`MOnly patients with recurrent disease (17 = 714) are included in these analyses.
`
`domain of the HER-2/c-neu protein, but was 56%
`among women with low circulating levels [21]. How-
`ever,
`the decrement in efficacy experienced by pre-
`menopausal women in the present study is even more
`dramatic for droloxifene than for tamoxifen. It may be
`that the shorter serum half-life of droloxifene (1 day)
`as compared to tamoxifen (1 week), with the resulting
`fluctuations in serum droloxifene levels, could be an
`important factor. The kinetics of droloxifene in tumor
`tissue have not been examined, although this has been
`done for tamoxifen [13]. It is plausible that the ac-
`tive tamoxifen metabolite (4-OH-ta1noxifen) is able to
`
`compete more effectively than droloxifene for the es-
`trogen receptor in the face of endogenous estrogen, as
`would be the case in the pre-menopausal patient. One
`could speculate that a higher dose of droloxifene could
`have overcome this disadvantage, and there is some
`evidence for this. A series of investigations carried out
`to study the effects of tamoxifen and droloxifene on
`serum levels of sex hormones, IGF-1 and associated
`binding proteins in breast cancer patients [14—18] has
`shown that these effects are dose related for drolox-
`
`ifene between 40 mg/d, the dose in the present study,
`and 100 mg/d.
`In addition, other data obtained in
`
`AstraZeneca Exhibit 2083 p. 10
`
`

`

`Proportion
`.
`still
`alive
`
`1.0
`
`0.9
`
`0.8
`
`0.7
`
`0-6
`0.5
`
`0.4
`
`0.3
`
`0.2
`
`0.1
`
`0.0
`
`Phase III randomized trial of droloxifene versas tamoxifen
`
`171
`
`‘ ‘ “ —- .. ‘_
`
`\_
`
`—"L. _. .—
`"a _ ._ .. 1
`
`_ _I_
`
`L _______
`I
`
`
`
`Treatment group
`
`
`
`
`— Droloxifene
`— _ Tamoxifen
`
`
`Y
`I
`V
`l
`I
`l
`l
`I
`I
`I r I
`I
`l
`I
`l
`l
`I
`l
`I
`T
`I
`I
`l
`I
`I
`I
`l
`I
`I
`l
`I
`I
`I
`i
`I
`I
`l
`I
`l
`l
`I
`l
`I
`I
`l
`I
`I
`V
`I
`100
`200
`300
`400
`500
`600
`700
`800
`900
`1000
`
`0
`
`Figure 2. Kaplan—Meier curve — proportion still alive by treatment group (product limit estimator).
`
`Days
`
`normal postmenopausal women (unpublished obser-
`vations) demonstrate that the efficacy of droloxifene
`for other clinical endpoints (bone density, lipid lower-
`ing) does continue to increase substantially at doses
`beyond 40 mg. These effects are primarily estrogen-
`agonist effects. While no Phase 11 breast cancer trials
`with droloxifene have provided any convincing evid-
`ence that this is the case for anti-estrogen effects in
`breast tissue [1—8], these earlier trials, all much smal-
`ler than the present study, had little statistical power to
`discriminate between two active doses. Moreover, pre-
`menopausal women were not well represented in the
`Phase 11 breast cancer trials. Finally, given that these
`results show tamoxifen efficacy to be much lower in
`younger patients, one could speculate that a tamox-
`ifen dose higher than the standard 20 mg/day could
`increase efficacy in younger or premenopausal patients
`treated with tamoxifen.
`
`The wide geographic reach of this study provides
`some interesting insight into the nature of a global
`patient population, and their response to therapy. How-
`ever, it should be understood that the statistics seen
`here do not necessarily represent the epidemiology
`of breast cancer in these countries, but rather reflect
`the patients available and willing to participate in a
`research study. As such, they may be influenced by
`factors other than incidence,
`for example, cultural
`
`factors. Our findings do suggest that advanced disease
`is less commo