CLINICAL REVIEW
`
`Clinical Review Section
`
`
`
`
`
`92381L/ 0018 vhase 2 PK, efficac
`
`
`A Partially-Blind, Randomized, Multi-center Trial to Compare the Anti—Tumour Effects,
`Pharrnacokinetics and Tolerability of 50 mg, 125 mg and 250 mg Single Doses of
`
`
`FASLODEXTM (Long-Acting ICI 182,780) with Tamoxifen and with Tamoxifen Placebo
`
`
`in Postmeno.ausal Women Prior to Sur- e
`for Prima Breast Cancer.
`
`
`
`
`
`
`Conclusions: Whereas fulvestrant treatment resulted in a reduction in PgR index,
`
`
`tamoxifen caused an increase in the level of this protein, thus supporting the concept that
`tamoxifen and fulvestrant have different modes of action. Presumably fulvestrant exerts
`its effects by down-regulation of ER protein. At a dose of 250 mg, fileestrant also
`
`
`
`resulted in a statistical]
`si nificant ; eater decrease in ER index than tamoxifen.
`
`200 o-ostmenoausal women with urima
`breast cancer
`rim—m
`
`
`
`92381L/ 0039 Phase 1] PK efficac
`
`An Open, Randomized, Multi-center, Parallel-group Trial to Compare the Pharmacokinetics and
`Tolerability of 250 mg Single Doses ofFASLODEXTM given as a Single 5 ml or as Two 2.5 ml
`ln‘ections in Postmeno .ausal Women with Advanced Breast Cancer (923 BIL/0039
`Location
`
`Start/Sto- dates
`
`8/99-1/00 comleted
`
`rofile.
`
`Accrual
`
`desi n
`Ob'ectives
`
`38 ost memo-ausal women with advanced breast cancer
`
`Oen randomlzed narallel ; ou.
`PK, tolerabili
`
`Conclusions: There was no observed difference in the pharmacokinetics of a 250 mg
`dose of LA im fulvestrant following administration as either one 5 ml injection or as two
`2.5 ml injections. Fulvestrant, at a dose of 250 mg, was well tolerated when administered
`by either of the 2 methods, and the combined safety data from both treatment groups also
`demonstrated 3 00d safe
`
`ii.
`
`Phase 3 studies reviewed in detail
`
`92381L/ 0020 Phase 1]] efficac
`
`An Open, Randomized, Multi-center Trial Comparing the Efficacy and Tolerability of 125 mg and 250 mg
`of FASLODEXTM (Long—acting ICI 182,780) with 1 mg ARIMIDEX TM (Anastrozole) in Postmenopausal
`Women with Advanced Breast Cancer
`
`
`
`
`
`
`
`
`
`
`
`Ob'ectives
`PK, tolerabili
`, efficac , safe
`
`Page 38
`
`

`

`
`
`CLINICAL REVIEW
`
`Clinical Review Section
`
`92381L/ 0021 Phase II] efficac
`
`
`A Double-blind, Randomized, Multicenter Trial Comparing the Efficacy and Tolerability
`
`
`
`of 125 and 250 mg of FASLODEX (Long-acting ICI l82,780) With 1 mg of ARIMIDEX
`
`(Anastrozole) in Postmeno ausal Women With Advanced Breast Cancer
`North America multicenter
`
`Start/Sto- dates
`
`5/97-8/00 on Hoin for survival
`
`
`
`
`
`
`
`Accrual
`
`
`
`
`
`473 postmenopausal women with advanced breast cancer
`oro_ essed followin_ hormonal thera
`Phase III randomized double blind double dumm
`
`PK, tolerabili Efficac ,safe
`
`
`
`iii.
`
`Ongoing studies in first line indication
`
`Prelimina
`results discussed with A: alicant, trials not reviewed in detail.
`
`
`
`92381L/ 0025 Phase III efficac — first Line
`
`
`
`A Double-blind, Randomized, Multicenter Trial Comparing the Efficacy and Tolerability
`of 250 mg ofFASLODEX (Long—acting ICI 182,780) with 20 mg ofNOLVADEX
`(Tamoxifen) in Postmeno-ausal Women With Advanced Breast Cancer
`'
`Location
`North America multicenter
`
`Start/Sto dates
`
`5/97-8/00 onoin for survival
`
`Accrual
`
`desi ;
`
`Ob'ectives
`
`473 postmenopausal women with advanced breast cancer
`ro_ressed followin hormonal therao
`
`Phase II] randomized double blind
`
`Efficac , safe
`
`Preliminary Conclusions: Time to progression in patients treated with Faslodex was
`inferior to TTP in uatients treated with Tamoxifen — 206 da 5 vs 252 da 5 for Tamoxifen.
`
`Reviewer comment: although the TTP results for the first line indication appear to be inferior
`for Fulvestrant compared with Tamoxifen, after internal discussion these results were not
`considered to affect conclusions regarding the results oftrials in the second line indication. (see
`appendix 2 for more complete discussion oftrial #25)
`
`c.
`
`Detailed Review of Trials by Indication
`
`The descriptions in this section are based on the Applicant’s Trial Protocol submitted to the
`NDA.
`
`i.
`
`Proposed indication
`
`Page 39
`
`

`

`
`
`CLINICAL REVIEW
`
`Clinical Review Section
`
`The Applicant’s proposed indications is " “.2
`,~_“"
`“M“
`
`wrmu1:‘w~l4
`
`ii.
`
`Overview of the clinical trial program
`
`(1)
`
`Summary
`
`The fulvestrant clinical trial program comprises 22 completed or closed trials and 4
`ongoing trials, with 1877 subjects exposed to trial treatment (including fulvestrant, anastrozole,
`tamoxifen, goserelin acetate, or placebo) as of the last data-cutoff date (30 June 2000). Efficacy
`end points were evaluated only in trials in which patients received the LA im formulation,
`specifically, in Trial 0004, the Phase II efficacy trial, and in Trials 0020 and 0021, the Phase [[1
`controlled trials designated as pivotal for this submission. Of the 1877 subjects enrolled in the
`clinical trial program, 1014 (54%) patients from 166 centers in North America, Europe,
`Australia, and South Africa were randomized to treatment in the pivotal efficacy trials, with data
`from 851 included in the primary efficacy analyses. All patients were included in the evaluations
`of safety and tolerability.
`
`(2)
`
`Selection of comparator agent
`
`The applicant cited several reasons for selection of anastrozole as the comparator agent in both
`phase 3 efficacy trials. Anastrozole produces known objective response rates comparable to or
`better than that of megestrol acetate, the progestin most commonly used as comparator in
`previous registration trials in the second line setting. In clinical trials, objective response rates
`with anastrozole reached 10.4% when given as second-line therapy, compared with 5.5% and
`10.4% with megestrol acetate. Additionally, anastrozole is well tolerated and does not induce the
`typical steroid-like side effects seen with progestins. Wide acceptance and use among physicians
`as an effective treatment of advanced breast cancer in postmenopausal women with disease
`progression afier tamoxifen therapy was also cited. The FDA agreed that anastrozole was an
`acceptable comparator for both trials, although we did suggest consideration of the use of
`megestrol as comparator in one of the trials.
`
`Reviewer Comment: Previous second line approvals in advanced breast cancer have been based
`on randomized non inferiority trials against the progestin agent megesterol acetate 160 mg/d in
`patients who have progressed after treatment with tamoxifen. Anastrozole was approved after
`review of 2 phase 3 trials in 764 patients with similar entry criteria as the present NDA. The
`primary endpoints were response rate and time to progression, as in the current trials under
`review. These trials initially compared the aromatase inhibitor anastrozole against 2 doses of
`fulvestrant, a the selective estrogen receptor modulator. The trials are therefore of very similar
`design compared with the previous registration trials and" with each other, except that 0021 was a
`double blind, double dummy and 0020 was an open label design. The trial plans and efficacy
`data will therefore be reviewed concurrently, and a few minor differences will be noted. The
`original design was based on achievement of statistical superiority in time to progression. In
`
`Page 40
`
`

`

`
`
`CLINICAL REVIEW
`
`Clinical Review Section
`
`retrospect this may not have been realistic, since no previous registration trial had been able to
`demonstrate superiority of time to progression in the second line treatment of metastatic breast
`cancer.
`
`iii.
`
`Phase 3 clinical trials
`
`(1) Overview
`
`The Phase III clinical trial program comprised 2 controlled trials, Trials 0020 and 0021.
`Both trials were multicenter, randomized, parallel-group trials with patients receiving IM
`fiilvestrant (125 or 250 mg monthly) or oral anastrozole (1 mg daily). Trial 0020 was conducted
`in Europe, Australia, and South Africa, and Trial 0021 was conducted in North America. In
`Trial 0020, treatment was open label, and fulvestrant 250 mg was administered as a single 5-ml
`im injection. In Trial 0021, treatment was double-blind (double-dummy approach), and
`fulvestrant 250 mg was administered as two 2.5-m1 serial im injections (1 per buttock). Each
`trial compared the efficacy and safety of fulvestrant with that of anastrozole.
`
`(2)
`
`Trial 92381L/0020: European Trial
`
`(a)
`
`Title
`
`An Open, Randomised, Multicentre Trial Comparing the Efficacy and Tolerability of 125
`mg and 250 mg FASLODEXW(Long-Acting1C1 182,780) With 1 mg of ARIMIDEX‘“
`(Anastrozole) in Postmenopausal Women With Advanced Breast Cancer.
`
`(b)
`
`Summary of trial design
`
`Trial 0020 was an open-label, randomized, parallel-group, multicenter trial conducted in 83
`centers in Europe, Australia, and South Africa. The design was essentially identical to trial 0021,
`except that it was an open-label trial and fulvestrant 250 mg administered as a 5-ml injection
`instead of two 2.5-ml injections as in Trial 0021 in accordance with European guidelines which
`differ from US guidelines concerning intramuscular-injection volumes. Initially, patients who
`met the eligibility criteria were allocated to the following randomized treatments on a 1:121
`basis: either
`
`a)
`
`fulvestrant 125 mg (2.5 ml) im monthly or
`
`b)
`
`fulvestrant 250 mg (5 ml) im monthly, or
`
`c) anastrozole 1 mg po daily
`
`Patients continued treatment until objective disease progression or other events required
`withdrawal; at such time, trial treatment was stopped, and standard therapy was initiated;
`thereafter, patients were followed up until death to determine survival interval. Patients who
`
`Page 41
`
`

`

`
`
`Clinical Review Section
`
`withdrew from trial treatment before progression were followed up until objective disease
`progression and death.
`
`(3)
`
`Trial 92381L/ 0021: North American Trial
`
`(a)
`
`Title
`
`A Double-blind, Randomized, Multicenter Trial Comparing the Efficacy and Tolerability
`of 125 and 250 mg of FASLODEX (Long-acting lCl 182,780) With 1 mg of ARIMIDEX
`(Anastrozole) in Postmenopausal Women With Advanced Breast Cancer.
`
`(b)
`
`Summary of trial design
`
`This was a double-blind, randomized, multicenter, parallel-group trial. This trial compared the
`efficacy and safety (tolerability) of fulvestrant injections with that of oral anastrozole and
`assessed the pharmacokinetics of fulvestrant following injection of the LA im formulation.
`Initially, patients who met the eligibility criteria were allocated to the following randomized
`treatments on a l:l:l basis: either
`
`a)
`
`fulvestrant 125 mg (2.5 ml) im monthly plus anastrozole placebo po, daily or
`
`b)
`
`fulvestrant 250 mg (2x25 ml) im monthly plus anastrozole placebo po od, or
`
`c) anastrozole 1 mg po daily plus placebo 2x25 ml im monthly
`
`(4) Design aspects common to both trials:
`
`(a)
`
`Treatment plan (initial)
`
`Trial 20
`
`Trial 21
`
`
`————
`
`
`i.m.monthl
`
`i.m. month]
`
`dail
`
`month]
`
`dail
`
`dail
`
`dail
`
`i.m. monthl
`
`(b)
`
`Major Protocol amendments
`
`There were 2 amendments to the protocol. The first occurred after 30 patients randomized to
`treatment with fulvestrant 125 mg (across trials) had been treated and monitored for 3 months.
`The responses were assessed, the protocol was subsequently revised, and the 125—mg treatment
`group was discontinued from this trial. Initially, a total of 588 patients (196 patients per each of
`the three treatment groups) were to be recruited over a 24-month period, with a minimum
`
`Page 42
`
`

`

`
`
`CLINICAL REVIEW
`
`Clinical Review Section
`
`follow-up period of 6 months. When 490 subjects either showed disease progression or died
`(end-point events), the 3 treatment groups were to be analyzed and compared for efficacy and
`tolerability.
`
`Effective 27 April 1998, the primary objective was changed to comparing the effect of LA
`fulvestrant (250 mg im) with oral anastrozole (1 mg daily) in terms of time to progression in
`postmenopausal women with advanced breast cancer.. The number of patients required was
`changed from 588 to 392. Recruitment would continue until 340 events from the remaining 2
`groups had occurred. When 340 subjects either showed disease progression or died (endpoint
`events), the treatment groups were to be analyzed and compared for efficacy and tolerability.
`
`Reviewer comment: This three arm design was also used in two other registration trials in this
`indication: anastrozole and letrozole. This design allowed for the determination ofa dose-
`response and optimization of the dose. The initial analysis was not considered an interim analysis
`and no statistical adjustment was necessary. The trial essentially restarted when the 125mg group
`was dropped.
`
`The second amendment, effective 24 September 1999, redefined the statistical methods used to
`analyze quality-of-life data.
`
`(c)
`
`Inclusion criteria relating to indication:
`
`0
`
`histologic or cytologic confirmation of breast cancer
`
`0 Objective evidence of recurrence or progression of disease not considered amenable to
`curative treatment
`
`0
`
`0
`
`postmenopausal woman, defined as any of the following:
`> age greater than or equal to 60 years
`> age greater than or equal to 45 years with amenorrhea for longer than 12 months and an
`intact uterus,
`> follicle-stimulating hormone (FSH) levels within the postmenopausal range (utilizing
`ranges from the testing laboratory facility),
`> having had a bilateral oophorectomy
`relapse after adjuvant endocrine therapy with an antiestrogen or a progesterone and no more
`than 1 prior hormonal therapy for breast cancer with second-line hormonal treatment or
`disease progressed after either an antiestrogen or progesterone as first-line treatment for
`advanced disease
`
`0 Evidence of hormone sensitivity, defined as
`> at least 12 months of adjuvant hormonal treatment before relapse, or
`> tumor remission or stabilization resulting from hormonal therapy for at least 3 months
`before progression in advanced disease,
`> estrogen—receptor positive (ER+) status or
`> progesterone-receptor positive (PgR+) status
`presence of at least ] measurable or evaluable lesion.
`> Measurable is defined as
`
`0
`
`/ clinically measurable in 2 perpendicular axes
`
`Page 43
`
`

`

`
`
`CLINICAL REVIEW
`
`Clinical Review Section
`
`'\
`
`at least ] dimension greater than or equal to 2.5 cm or both dimensions greater than or equal to 1
`cm.
`
`Evaluable is defined as
`
`\'\‘\'\V
`
`both dimensions less than 2.5 cm (measured clinically) or 1 dimension less than 1.0 cm
`measurable in 1 axis only,
`not measurable but visible by photography,
`ssessable but not measurable by radiographic imaging (mediastinal lymph nodes or diffuse
`pulmonary infiltration, or osteolytic bone lesions)
`0 World Health Organization performance status of 0, 1, or 2
`
`(d)
`
`Exclusion criteria
`
`0
`
`presence of life-threatening metastatic visceral disease, extensive hepatic, CNS (past or
`present) or symptomatic pulmonary lymphangitic spread
`previous treatment for breast cancer with fulvestrant, anastrozole, or any aromatase inhibitor
`treatment with LH-RH analogs within 3 months before randomization
`
`more than 1 prior endocrine medical treatment for advanced breast cancer
`
`estrogen replacement therapy within 4 weeks before randomization
`
`The following are considered to be neither measurable or evaluable:
`> Lesions in previously irradiated fields
`> Diffuse lesions such as lymphedema, hilar enlargement, pleural effusion, ascites,
`metastases in the central nervous system, bone marrow infiltration, osteoblastic bone
`lesions, and osteolytic bone lesions
`
`Reviewer comment: Patients were not stratified. Irnbalances in previous treatment, hormone
`receptor status, or performance status could potentially have influenced results, and these
`population characteristics will be reviewed. The definition of ER/PR positivity was not provided.
`
`(e)
`
`Screening and clinic visits
`
`Screening: Within 3 weeks before randomization, baseline assessments were to be performed
`including: Medical history, Concomitant therapy, Demographic data, Concurrent conditions,
`Physical examination, Electrocardiogram, Hematology testing, Prothrombin time, Biochemistry
`testing, Chest radiograph, Isotopic bone scan or skeletal survey, Tumor assessment, QOL
`measurements. Table 15 lists the study plan and events:
`
`APPEARS THiS WAY
`ON ORlGlHAL
`
`Page 44
`
`

`

`
`
`Clinical Review Section
`
`Table 14: timing of events and assessments
`
`assessments
`
`Monthlyxa Everyamomhsxs
`
`
`
`
`
`
`Concomitant therapy
`
`Concurrent conditions
`Physical examination
`Blood pressure and pulse
`Weight
`Adverse events
`
`
`
`xxxxxxxxx
`
`xxxxx
`
`
`
`
`Hematology testingc
`
`Biochemistry testingdC
`
`
`
`efficac assessments
`Prima
`
`Tumor assessment —__
`
`Secondary efficacy assessments
`
`
`
`Analgesia use
`Global pain score
`
`
`Performance status
`
`
`
`
`
`
`
`xxxxxxxxx
`
`x><><><><><
`
`
`
`
`
`
`
`><
`
`><><><><
`
`><><><><><><
`
`Quality of Life
`Local site tolerance
`Health economics
`assessment
`
`
`(until progression)
`(within 3 days after randomization)
`(within 3 weeks before randomization)
`‘ Soft tissue
`masses assessed monthly x 3 months, objective radiological assessments q 3 months until progression)
`
`Reviewer comment: Since radiological assessments occur at three month intervals and soft
`tissue assessments occur monthly, an imbalance of measurable vs. evaluable patients could
`influence the time to progression. The necessity of fulvestrant patients to come into the clinic for
`injections might bias the progression endpoint against the fulvestrant aim in the open label trial.
`Clinic visits and trial plan seem otherwise adequate for safety and efficacy evaluation.
`
`Efficacy assessments
`
`(1) Objectives
`
`0 i
`
`v.
`
`(a)
`(b)
`
`Primary: time to disease progression
`Secondary
`(i)
`objective response rate
`(ii)
`time to treatment failure
`(iii)
`time to death
`(iv)
`duration of response,
`(v)
`symptomatic response,
`(vi)
`quality of life.
`
`(2)
`
`Primary end point : time to disease progression.
`
`0 Time to progression was defined as the time from randomization to disease progression.
`0
`The date ofprogression was defined as follows:
`
`Page 45
`
`

`

` CLINICAL REVIEW
`
`Clinical Review Section
`
`> If for a measurable lesion the date when there was a documented increase in the size
`
`(product) of more than 25% compared with the minimum dimensions recorded.
`> The date progression was documented by the investigator from an evaluable lesion.
`> The date a new lesion was noted or progression was documented from an additional
`lesion or breast-cancer related condition.
`
`> If the patient died, the date of death from any cause
`
`0 Patients whose disease had not progressed at the time of data cutoff were right censored
`using the date of last assessment.
`
`(3) Objective tumor response
`
`Tumor response was a secondary endpoint, but became a co primary endpoint following revision
`of the objectives from demonstration of superiority in TTP to non inferiority in response rate and
`TTP. Assessment of tumor response was made for both measurable and nonmeasurable disease
`and involved assigning response categories to previously identified lesions or tumors. For each
`selected clinically measurable lesion, 2 dimensions (length and width) were recorded.
`
`(a)
`
`UICC response criteria
`
`The categories of objective tumor response assigned at each visit were defined according to
`standard UICC (Union Internationale Contre le Cancer) criteria:
`
`(i)
`
`Complete reSponse
`
`0 No clinical or radiological evidence of residual lesions on 1 visit, with no evidence of disease
`recurrence or death within 4 weeks of response assessment.
`0 For patients with evaluable disease of the bone only, the following were required
`remineralization of all lytic lesions with radiological evidence of calcification,
`absence of bone pain (without analgesics),
`no new pathological fractures within 4 weeks of the assessment,
`evidence of bone remodeling in previously distorted bone, and
`normalization of bone as determined from bone scan.
`
`VVVVV
`
`(ii)
`
`Partial response
`
`A PR was recorded when disease improved (compared with baseline assessment) on 1 visit, and
`disease progression was not evident, based on the following:
`0
`for measurable disease, a decrease of at least 50% in the sum of the products of the 2
`largest perpendicular diameters of all the measurable lesions without
`>
`(a) an increase of more than 25% in the size of any lesion or
`>
`(b) the appearance of any new lesion
`
`0
`
`for nonmeasurable (evaluable) disease, objective improvement on the basis of radiological,
`ultrasonic, or photographic evidence.
`
`Page 46
`
`

`

` CLINICAL REVIEW
`
`Clinical Review Section
`
`For patients with evaluable disease of the bone only, partial remineralization of lytic lesions
`without new pathological fractures or new bone lesions.
`
`(iii)
`
`Stable disease
`
`Lack of objective disease progression and insufficient evidence for designating either
`complete or partial objective response
`Stable disease was further categorized as either greater or less than 24 weeks (168 days)
`
`(iv)
`
`Disease progression
`
`any measurable lesion increased in size (product of the 2 largest perpendicular diameters) by
`more than 25% compared with the minimum dimensions recorded during the trial,
`
`existing lesions worsened (determined from radiological, ultrasonic, or photographic
`evidence or clinical assessment) or
`
`new lesions appeared
`
`(b)
`
`Assignment of response category by computer
`algorithm in patients with measurable disease only
`
`For patients with measurable disease only, objective response was assigned using a computer
`algorithm based on UICC response categories. The best response determined by the computer
`algorithm was used in the primary statistical analysis:
`
`Yrs
`
`Yet
`
`Any new lemons?
`No
`
`Compare and vs smallest prrvioux
`In: for each lesion
`
`Any area >[l15 : miles pix-“ms nu)?
`
`STOP, Response = Progression
`
`Na
`
`STOP. Rcspmac = Rupam: II pinion: vim
`
`Yes
`
`Na
`
`Yes
`
`a
`
`I|z
`
`An- :ll lumps “lewd"
`
`Y‘s
`
`All fictions = Zero?
`
`No
`
`STOP; Response = Convict: response
`
`Camera total are: ofall ksmm
`with ml ma u balm:
`
`STOP. Rts'pomc = Sub]: discxsc
`
`No
`
`Totll mm (0.5 x tmal urea at baseline)?
`Yes
`
`Page 47
`
`510?: Rcspomc = Partial response
`
`

`

`
`
`CLINICAL REVIEW
`
`Clinical Review Section
`
`(c)
`
`Assignment of global response by investigator
`
`A best response of CR or PR was assigned when a patient satisfied the respective criteria for
`response on one visit occasion, with no evidence of disease recurrence or death within 4 weeks
`of response assessment. The investigator’s global overall assessment was also used for a patient
`with only measurable disease if a bi dimensional lesion became uni dimensional. If progression
`was assigned as an overall visit objective tumor response, all subsequent visit response
`assessments were considered progression. For each patient, best objective response over time
`was determined on the basis of objective response assessment per visit.
`
`Reviewer comment: The design of the randomized fulvestrant registration trial was similar to
`the design of the anastrozole registration trial. UICC response criteria were also used in the
`previous anastrozole NDA registration trials.lz These required 2 dimensional measurements,
`while the newer RECIST criteria allow single measurements to be included as measurable
`lesions. There are several differences in definition of partial responses and progressive disease,
`however comparisons between response data assessed by the two techniques have shown
`comparable response assessments. 3 Therefore, use of the older UICC/WHO efficacy criteria
`were unlikely to influence the validity of efficacy results. Since final assignment of responses
`was determined by the investigator in those patients without ‘measurable disease only,’
`investigator bias might potentially have affected the results of the open label European trial
`(#20).
`
`(4)
`
`Survival
`
`Survival was defined as the number of days from randomization until death.
`Survival status of patients was recorded every 3 months after disease progression or after
`withdrawal for any reason until death. Patients still alive at the time of data cut-off were
`censored to the last date they were known to be alive.
`
`(5) Duration of response
`
`Duration of response was calculated only for patients who had best responses of CR or PR.
`Duration was defined in two ways: (I) as the number of days from randomization until the day
`
`'2 Union Internationale Contre la Cancer response criteria were published in 1977 by Hayward
`et. a]. in the European Journal of Cancer, 13:89-94, and in the 1979 WHO Handbook for
`reporting results of cancer treatment. Concerns regarding the reproducibility and clinical
`applicability ofthe WHO response criteria led to the proposal by James et. al. in JNCI 91:523-
`526 (1999) for the simplification of response evaluation through the use of unidimensional
`measurements and the sum of the longest diameters instead of the WHO method using two
`measurements and the sum of the products. See Therasse P, et. al New Guidelines to Evaluate
`the Response to Treatment in Solid Tumors, JNC] 922205-216, 2000 for a recent discussion.
`'3 For a comparison between WHO and RECIST criteria, see Appendix
`
`Page 48
`
`

`

`
`
`CLINICAL REVIEW
`
`Clinical Review Section
`
`on which disease progression was first observed and (2) as the number of days from the date that
`objective response was first documented until the date of disease progression.
`
`(6)
`
`Time to treatment failure
`
`Time to treatment failure was defined as the number of days from randomization until the
`earliest occurrence of disease progression, death, or withdrawal from trial treatment for any
`reason. The date of treatment withdrawal was defined as the date on which a patient or her
`physician decided to discontinue trial treatment.
`
`(7)
`
`Symptomatic response
`
`Patients were assessed over time for change in analgesic use, global pain, and WHO
`performance status or level of daily activity.
`
`(8) Quality of Life analyses
`
`(a)
`
`Treatment outcome index (TOI)
`
`The main QOL variable was the treatment outcome index (TOI). The T0] was created
`from the FACT-B QOL questionnaire and reflects the sum of scores for the following
`subscale dimensions: functional well being (questions 27 to 33), physical well being
`(questions 1 to 7) and breast cancer concerns (questions 35 to 43). In the formal statistical
`analysis, the difference in T01 over time between fulvestrant 250 mg and anastrozole was
`compared using a random coefficients model for the intent-to-treat population only. The
`model was adjusted to account for both within-patient (due to repeated measures) and
`between-patient
`information and included terms to account for baseline TOI score and
`baseline covariates.
`
`(b)
`
`Time to deterioration in QOL
`
`Time to deterioration in QOL, a secondary variable, was defined as the time between
`randomization and the earliest occurrence of a 5-point reduction in the T01 from baseline, or
`death. If a patient had not died or did not have the 5-point reduction in T01 at their last QOL
`assessment, then this observation was right-censored at the time of the last assessment.
`
`Reviewer cement: QOL analyses are often hampered by incomplete data collection. Labeling
`claims will have to be subject to scrutiny with attention to data collection.
`
`v.
`
`Statistical plan
`
`(1)
`
`Population
`
`(i)
`
`Intention to Treat Primary Analyses
`
`Page 49
`
`

`

`
`
`CLINICAL REVIEW
`
`Clinical Review Section
`
`The primary analysis was an intent-to-treat (H‘T) analysis and, as such, included data from all
`randomized patients. Treatment effects were compared on the basis of randomized treatment,
`regardless of treatment actually received, with data adjusted for baseline effects.
`
`(ii)
`
`Secondary Analyses
`
`Secondary analyses were conducted for a subset of patients who did not significantly violate or
`deviate from the protocol, the per-protocol population, by treatment received, and for the ITT
`population with data unadjusted for baseline effects.
`
`Reviewer comment: In a superiority trial the primary analysis is performed in the ITT
`population because it tends to avoid over-optimistic estimates of efficacy resulting from a per—
`protocol population, since the non-compliers included in the ITT population will generally
`diminish the estimated treatment effect. However, in a non-inferiority trial analysis results from
`the ITT population is generally not conservative; hence, analyses in both populations are
`equally important for a robust interpretation.
`
`(2)
`
`Sample size
`
`The estimation of sample size was based on the primary end point of time to progression. In
`previous anastrozole trials the median time to progression was 140 days. To detect a hazard ratio
`of greater than or equal to 1.43 or less than or equal to 0.70, at a significance level of 5% with
`90% power, 490 end point events (disease progression or death before progression) had to occur
`in each trial. Given that both trials had an estimated accrual time of 24 months, with 6-month
`
`follow-up periods, patient requirements were 196 patients per treatment group per trial or at least
`588 patients per trial. When the 125-mg treatment group was dropped, 196 patients would still
`be required in each of the remaining 2 groups for a total of 392 patients per trial. The analysis
`would be performed when at least 340 events occurred across the remaining 2 groups.
`
`(3)
`
`Interim Analyses
`
`(a)
`
`Initial Interim Analyses of 125 mg arm
`
`The first review of data occurred after 30 patients randomized to fulvestrant 125 mg (across
`Trials 0020 and 0021) were treated and monitored for a minimum of 3 months. The objective
`was to determine whether the 125-mg dose produced adequate evidence of clinical activity. If
`no response was seen, then treatment at this dose would be discontinued. At the time of the
`review, 1 (3.3%) patient had withdrawn, 9 (30%) had stable disease, and 20 (66.7%) had disease
`progression. Since the criteria for continuing treatment at this dose were not met, the 125-mg
`dose group was dropped from both trials. In conjunction with that decision, the DMC conducted
`a blinded review of data from the remaining two treatment groups and recommended that the
`trial continue as provided for in the protocol.
`
`Page 50
`
`

`

`
`
`Clinical Review Section
`
`(b)
`
`Planned Interim Analyses
`
`Interim analyses were conducted in both trials after 170 progression events were recorded across
`the remaining 2 treatment groups (per trial) to confirm the activity of fulvestrant 250 mg
`compared with that of anastrozole. Both time to progression and objective response rate were
`evaluated. Statistical analysis was applied to time to progression, with the nominal O’Brien
`Fleming significance level set at 0.4% for a two sided test (or equivalently, at 0.2% for a one-
`sided test). Descriptive statistics were used to summarize objective response data. The DMC
`reviewed the results of the two interim analyses and again recommended that the trials continue.
`
`(c)
`
`Final Analysis
`
`(i)
`
`TTP Superiority analysis (pre specified)
`
`For both trials, trial design and analysis were geared toward assessing whether fulvestrant was
`superior to anastrozole. Final analyses were conducted after 340 progression events across the
`remaining 2 treatment groups (per trial) were recorded. The nominal level of significance was
`adjusted from 5% to 4.86% because of the interim data summary and analysis. The pre specified
`response rate analysis was by adjusted logistic-regression model with baseline covariates: age,
`performance status, measurable compared with non-measurable disease, receptor status, previous
`response to hormone therapy, previous use of cytotoxic chemotherapy, and use of
`bisphosphonate therapy for bone disease. The prespecified time to progression analysis was by
`cox proportional-hazards model with baseline covariates: age, performance status, measurable
`compared with non-measurable disease, receptor status, previous response to hormone therapy,
`previous use of cytotoxic chemotherapy, and use of bisphosphonate therapy for bone disease.
`
`(ii)
`
`Applicant's Non inferiority analysis of
`Response rate and TTP (retrospective)
`
`When the superiority objectives were not met, the trials were retrospectively assessed for non-
`inferiority, compared with anastrozole, for the efficacy end points of time to progression,
`objective response, and time to treatment failure. For the analyses of time to progression and
`objective response rate, the applicant used a one-sided confidence interval of 97.57% (because of
`the interim analysis).
`
`(iii)
`
`FDA non inferiority Analysis
`
`The Division agreed with the applicant that, based on regulatory precedent in the second line
`hormonal treatment of advanced breast cancer, demonstration of non inferiority could provide
`the basis for marketing approval. The nominal significance levels pre-specified in the protocol
`were calculated based on an unadjusted logrank test with a pre-specified correlation between the
`two statistics (one for the interim and the other for the final analysis). The FDA statistical
`reviewer did