European Journal ofCancer l996;32A(3):404—412
`
`AN 403 91
`
`Original Paper
`
`A Randomised Trial Comparing Two Doses of the New
`Selective Aromatase Inhibitor Anastrozole (Arimidexf‘ With
`Megestrol Acetate in Postmenopausal Patients With' Advanced
`Breast Cancer
`
`W. Jonat,‘ A. Howell,2 C.‘ Blomqvist,3 W. Eiermann,‘ G. \Winblad,5 C. Tyrrell,6 L. Mauriac,7
`H. Roche,8 S. Lundgren,° R. Hellmundlo and M. Azablo on behalf of the ARIMIDEX Study
`Group
`
`1University Women’s Hospital, Eppendorf, Hamburg, Germany; 2The Christie Hospital, Manchester, UK;
`3University Central Hospital, Helsinki, Finland; 4Women’s Hospital, Munich, Germany, 5Department of
`General Oncology, Soder Hospital, Stockholm, Sweden; 6Plymouth General Hospital, Plymouth, U.K.; 7Institut
`Bergonie, Bordeaux, France; 8Centre Claudius Regaud, Rue de Pont Saint-Pierre, Toulouse, France; ”Regional
`and University Hospital, Trondheim, Norway; and mlkiledical Research Department, ZENECA Pharmaceuticals,
`Alderley Park, Macclesfield, Cheshire, U.K.
`
`
`The aim of this study was to compare the efiicacy and tolerability of the new aromatasc inhibitor
`‘ARIMIDEX’ (anastrozole) with megestrol acetate in the treatment of advanced breast cancer in
`postmenopausal women. Anastrozole is a new potent and highly selective non-steroidal aromatase
`inhibitor. We conducted a prospective randomised trial comparing two doses of anasn-ozole (l and
`10 mg orally once daily) with megestrol acetate (40 mg orally four times daily) in postmenopausal
`patients with advanced breast cancer who progressed after prior tamoxifen therapy. All patients were
`analysed for efiicacy as randomised (intention to treat) and for tnolerability as per treatment received.
`0f the 378 patients who entered the study, 135 were randomised to anastrozole 1 mg, 118 to anasu'ozole
`10 mg, and 125 patients to megestrol acetate. After a median follow-up of 192 days, response rate which
`included complete response, partial response and patients who had disease stabilisation for 6 months
`or more was 34% for anastrozole 1 mg, 33.9% for anastrozole 10 mg and 32.8% for megestrol acetate.
`There were no statistically significant differences between either dose of anastrozole and megestrol
`acetate in terms of obiective response rate, time to objective progression of disease or time to treatment
`failure. The three treannents were generally well tolerated, but more patients on megestrol acetate
`reported weight gain, oedema and dyspnoea as adverse events while more patients on anastrozole
`reported gastro—intestinal disorders, usually in the form of mild transient nausea. Patients on anastro-
`zole did not report higher incidences of oestrogen Withdrawal symptoms. Anastrozole is an efi'ective
`and well tolerated treatment for postmenopausal patients with advanced breast cancer. The higher
`10mg doSe did not result in additional clinical benefit, but was well tolerated reflecting the good
`therapeutic margin with anastrozole. Based on this data, anastrozzole 1 mg should be the recommended
`therapeutic dose.
`
`Key words: aromatase, postmenopause, breast neoplasms, comparative study, megestrol, phase III
`clinical trials, random allocation
`Eur] Cancer, Vol. 32A, No. 3, pp. 404—412, 1996
`
`'
`
`AstraZeneca Exhibit 2138 p. 1
`InnoPharma Licensing LLC V. AstraZeneca AB IPR2017-00904
`Fresenius-Kabi USA LLC V. AstraZeneca AB IPR2017-01910
`
`

`

`Arimidex Versus Megestrol Acetate in Breast Cancer
`
`405
`
`‘
`
`INTRODUCTION
`
`AFI‘ER TmoxrrnN, progestins and the aromatase inhibitors are
`currently among the commonly used endocrine agents for the
`treatment of advanced breast cancer in postmenopausal women
`In several randomised trials, these agents achieved similar effi-
`cacy to the anti—oestrogen tamoxifen [1—5] . However, progestins,
`such as megestrol acetate, are associated with a high incidence
`of weight gain, oedema and occasionally cardiovascular and
`thrombo-embolic side—effects
`[4,6,7]. Aminoglutethimide, a
`nonselective aromatase inhibitor, is associated with a high inci~
`dencc' of side—effects such as lethargy and rash and is often given
`alongside corticosteroid supplementation [2,8,9]. These side—
`efiects have restricted the use ofprogestins and aminoglutethim~
`ide to second- and third-line treatments following tamoxifen.
`» Approximately one third of human .breast cancers are oes-
`trogen~dependent and will regress following oestrogen depri-
`vation [10]. In postmenopausal women, the major mechanism
`for oestrogen production is the peripheral conversion (by
`aromatase) of the adrenal steroid androstenedione to oestronc -
`and subsequently to oestradiol [l 1]. In addition to peripheral
`aromatase activity, it is known that- about two thirds of breast
`tumours show aromatase activity which appears to provide a
`local source of oestrogens within the breast turnout [12], and
`oestrogen levels are higher in breast tumour than in plasma
`[13]. It is, therefore, theoretically possible that high doses of
`an aromatase inhibitor, which .could achieve higher tissue
`concentration of drug, might block oestrogen synthesis within
`the tumour more efficiently. To date, it has not been possible
`to adequately test this hypothesis, probably because lack of
`selectivity and poor tolerability of aminoglutethimide have
`limited its investigation to a relatively small range of doses of
`250~l 000 mg daily, with or without hydrocortisone [14—16].
`Anastrozole (Arirnidex) is an achiral benzyltriazole deriva-
`tive which has been shown to be a potent and highly selective
`aromatase inhibitor in preclinical and phase I clinical studies
`[17]. At doses of 1 mg daily and higher, anastrozole sup—
`pressed oestradiol to the maximum degree measurable. Doses
`up to 10 mg daily were investigated in early studies and did
`not have any effect on glucocorticoid or mineralocorticoid
`secretion as indicated by normal responses to ACTH stimu—
`lation tests [17]. In addition, anastrozole is rapidly absorbed,
`with maximal plasma concentration occurring within 2 h of
`oral administration, and has a long elimination half-life of 30—
`60 h allowing once daily dosing [17].
`This report describes a prospective randomised trial which
`investigated the efficacy and tolerability of two blinded doses
`of anastrozole (1 mg and 10 mg orally once daily) compared
`with that of megestrol acetate at its recommended therapeutic
`dose of 160mg daily (40 mgx 4 daily) in the treatment of
`postmenopausal women with advanced breast cancer. Anas—
`trozole 1 mg daily was the lowest dose producing maximal
`oestradiol suppression, while the 10 mg dose was the highest
`dose investigated in early clinical studies and which still.
`showed selectivity and good tolerability; the use of anastrozole
`10 mg dose offered the opportunity of achieving increased
`intratumour concentrations of the drug and hence providing
`more efficient aromatase inhibition at the tumour level, with
`potential additional clinical benefit.
`
`PATIENTS AND METHODS
`Pau'em population
`Patients were eligible for this study if they were postmenos
`pausal women progressing on first—line
`tamoxifen for
`
`advanced breast cancer, or if they were relapsing whilst either
`receiving or having completed adiuvant tamoxifen treaunem.
`Postmenopausal women were defined as being >50 years with
`no menstruation for the last 12 months or who have castrate
`levels offollicle stimulating honnone (FSH) (>40 IU/l). Since
`tamoxifen and its active metabolites have a very long half-life
`and can remain in the cells for l or more months, it was
`considered not feasible to require a tamoxifen withdrawal
`period in practice as many patients would be unwilling to
`withhold active therapy after the development of progressive
`disease on tamoxifen: For patients who were known to be
`oestrogen receptor (ER) negative, prior evidence of response
`to endocrine therapy ,was required. For all other patients, no
`minimum period of adjuvant tamoxifen or prior response to
`tamoxifen was required.
`Patients were excluded if they had life-threatening Visceral
`disease, (extensive hepatic involvement or any degree of cranial .
`or leptomeningeal spread or pulmonary lymphangetic spread),
`had previously been exposed to more than one cytotoxic
`chemotherapy regimen for advanced disease, or had received
`more than one prior hormonal treatment for advanced breast
`cancer. There was no upper age limit and performance status
`was World Health OrganisationtWHO) 0, 1, or 2. The study
`was approved by the ethics committee at each participating
`centre, and all patients gave written informed consent before
`enrolment. Patients with measurable lesions or evaluable but
`non-measurable lesions were eligible in the study. Patients
`with only blastic bone lesions were not considered to be evalu-
`able.
`
`Study design
`The. study was a phase III randomised trial with parallel
`group design; Eligible patients were randomised on a 1:1:1
`basis to receive orally either anastrozole 1mg once daily
`orlO mg once daily on a double-blind basis, or open-label
`- megestrol acetate 40 mg four times daily, Randomisation was
`effected centrally using a computer generated random scheme.
`Subjects were allocated to treatment in balanced blocks which
`were assigned on a centre basis. All randomised patients were
`followed up until progression andlor death.
`_ Systemic treatment for breast cancer, other than random-
`ised treatment, was not permitted until disease progression.
`Radiotherapy was allowed, but irradiated lesions were con-
`sidered non-evaluable for tumour response assessment unless
`for the assignment ofprogression.
`Baseline screening assessments were completed within the
`4 weeks prior to randomisation. On day l
`(the date of -
`randomisation), eligible patients underwent a complete physi-
`cal examination. Obiective assessments for, local and regional
`disease, together with biochemical, haematological and oestra—
`diol measurements were made at day 1, weeks 4, 8, 12, 16,
`20, 24 and every 12 weeks thereafter, until progression of the
`disease. Tumour evaluation included physical examination for-~
`superficial skin or soft tissue lesions, radionuclide bone scans,
`skeletal X-mys and chest X-rays for bone or pulmonary met—
`astases. Head and liver CT scans were performed if clinically-
`indicated.
`.
`-
`
`»
`
`Quality of life was: assessed using the Rotterdam Symptom.
`Checklist (RSCL) and a prospective subjective symptoms
`score. The RSCL was given to all patients for completion on
`day 1 and again every 12 weeks for 1 year, or until progression.
`The RSCL covers physical and psychological symptoms and
`the functional activity of the patient
`[18]. The subjective
`
`AstraZeneca Exhibit 2138 p. 2
`
`

`

`406
`
`,'
`
`.
`
`W. Jonat er al.
`
`symptom score was used at the same timepoints to evaluate
`analgesic use (ll—grade scale), bone pain, and WHO perform-
`ance status (S—grade scale each).
`In the evaluation of tolerability, any detrimental change in
`the condition of patients during the trial other than breast
`- cancer disease progression was recorded as an adverse event
`irrespective of causality. Adverse events were documented at
`each visit and monitored until they resolved.
`
`Endpoint:
`The primary endpoints were time to objective disease pro
`gression, objective response rate and tolerability. The second-
`ary endpoints were time to treatment failure, survival, duration
`of response, quality of life and subjective symptom scores.
`The assessment of tumour response included the evaluation
`of both measurable and evaluable non-measurable lesions.
`For measurable lesions, all measurements were source data
`audited and the measurements were then assessed by a valiv
`dated computer algorithm which assigned tumour response
`category based on per cent tumour regression applying a
`strict interpretation ofthe UICC (International Union Against
`Cancer) criteria of response [19]. Complete response was only
`assigned if all lesions disappeared; partial response was only
`assigned if at least 50% regression of the sum of all measurable
`lesions was achieved; any 25% or more increase from the
`minimum recorded size of lesions or appearance of a new
`lesions was assigned objective progression. The use of the
`computer programme to assign the response category for
`measurable lesions from recorded measurements was adopted
`to decrease potential subjective bias of individual investigators
`in assigning responses to study patients. For all patients with
`only evaluable non—measurable lesions,
`including patients
`with osteolytic bone lesions, the category of partial response
`was not allowed to provide a rigorous objective response
`assessment. For all patients, complete or partial responses had
`to be confirmed by two successive turnout assessments at least
`4 weeks apart. Response rate was calculated for all randomised
`patients (intention to treat basis) and nomevaluablc patients
`were, therefore, included in the denominator as non~responders.
`Time to progression, Lreauueirt failure, duration oftesponse
`and survival were calculated Erom the date of randomisation.
`Time to progression was the time until objective evidence of
`progression or until death from any cause, whichever occurred
`first. Time to treatment failure was the time until objective
`progression, death or treatment withdrawal for any reason,
`whichever occurred first. Duration of response was the time
`to progression for responding patients.
`
`Statistical analysis
`The study was sized on the basis of the primary endpoints
`of time to progression and tumour response rate. A minimum
`of 300 patients recruited at a uniform rate over 12 months
`with a minimum follow-up of 6 months was expected to
`provide 80% power to detect a treatment difference of
`approximately 14 weeks in median time to progression,
`assuming an overall median of 26 weeks, at the tWo—sided
`alpha = 0.05 significance level. This size of study was expected
`to provide 90% power to detect a treatment difference of
`approximately 20% in tumour response rates, assuming an
`overall response rate of 25%, at the two sided alpha 20.05
`significance level.
`An early interim analysis, included in the protocol, was
`conducted on primary endpoints only (time to progression
`
`and response rate). To allow for this, the O’Brien and Fleming
`method adjustment to the significance levels was used [20],
`and hence for the final analysis on these endpoints, signifi—
`cance was set at 4.8% level. In addition, for each endpoint,
`two analyses were Conducted—anastrozole 1 mg versus taegc_
`strol acetate, and anasu-ozole 10 mg versus megesrrol acctatc‘
`Therefore,
`in order to allow for this multiple testing, the
`level set for significance was then halved. Thus, the objective
`response and time to progression were assessed at the 2.4%
`level of significance, and other eflicacy endpoints including
`quality of life and subjective scores at the 2.5% level. All
`objective efficacy endpoints were analysed on the basis of
`the treatment to which the patients were randomly assigned
`(intention to treat), while tolerability analyses were conducted
`on the basis of treatment actually received by the patients.
`Time to treatment failure, time to disease progression and
`death were subjected to Cox’s Proportional HazardsModel.
`The proportion of responders (complete response‘and partial
`response) was compared between treatment groups, using
`logistic regression. The estimated treatment efi‘ect was pre—
`sented as an odds ratio with appropriate confidence intervals.
`For the Cox’s Proportional Harards Model and the logistic
`regression analyses, the hormone receptor status at entry and
`the disease status were used as covariates, and a test for
`treatment by covariate interaction was performed. RSCL
`scores were analysedby analysis of covariance (physical and
`psychological dimensions) and the Wilcoxon ranked-sum test
`(functional dimension). Subjective scores were analysed using
`logistic regression. Pharmacological adverse events, identified
`prospectively, were compared between treatments using Fish~
`er’s Exact Test. In addition, pairwise comparisons between
`anastrozole and megestrol acetate groups were performed on
`the number of patients with weight gain of at least 5% and at
`least 10% using Fisher’s Exact Test.
`
`RESULTS
`
`Study population
`A total of 378 postmenopausal advanced breast cancer
`patients from 73 centres were randomised into the trial
`between April 1993 and June 1994: 135 were randomised to
`receive anasttozole 1 mg once daily, 118 to receive anastrozolc
`10 mg once daily and 125 to receive megestrol acetate 40 mg
`four times daily. A total of 376 of the 378 randomised patients
`started study treatment. Patients’ baseline characteristics are
`shown in Table 1. There were no clinically significant imbal—
`ances in the three treatment groups. 'A slightly greater percent-
`age of patients on anastrozole 10 mg had experienced a prior
`response to tamoxifen for advanced disease, while a greater
`percentage of patients on anastrozole 1 mg had visceral
`lesions.
`'
`
`Efiicacy endpoint:
`'At the time of data cut-off for the analyses, the median
`duration of follow-up was 192 days for anasttozole l mg, 185
`days for anastrozole [0 mg and 182 days for megestrol acetate.
`There 'were no statistically significant difierenccs between
`anastrozole 1 mg or [0 mg daily and megestrol acetate with
`respect to median time to progression (132, 156 and 120 days,
`respectively), time to treatment failure (12], 128 and 115
`days, respectively) and survival (84.4%, 81.4% and 77.6%
`respectively). Figure 1a,b shows the Kaplan—Meier plots for
`time to disease progression and survival, respectively. Similar
`numbers of objective responses were observed in the three
`
`AstraZeneca Exhibit 2138 p. 3
`
`

`

`Arimidex Versus Megestrol Acetate in Breast Cancer
`
`407
`
`Table 1. Patients’ baseline characteristics
`
`Anastrozole
`Anastrozole
`1 mg/day
`10 mg/day
`(I: =135)
`(n = 118)
`
`
`Megestrol acetate
`40 mg four times
`daily (n : 125)
`
`WHO performance status
`0
`l
`2
`3
`
`51%
`37%
`12%
`0
`
`42%
`39%
`18%
`1%
`
`45%
`42%
`14%
`0
`
`65 (38—97)
`67 (444104)
`160 (140—176)
`
`66 (44787)
`67 (35—1 18)
`160 (135—178)
`
`64 (40-84)
`67 (45— 130)
`161 (143—175)
`
`Mean age and range (years)
`Mean weight and range fig)
`Mean height and range (cm)
`Prior therapy
`Adjuvant tamoxifen only
`Median disease free interval (months)*
`Tamoxifen for advanced disease
`Prior response to tamoxifen’r
`Prior chemotherapy
`Prior radiotherapy
`Receptor status
`ER+
`PR+
`ER and PR unknown
`Measurable disease
`No measurable disease
`Disease sites
`Soft tissue
`Bone
`Visceral
`Liver
`Disease extent
`20%
`19%
`11%
`Soft tissue only
`29%
`25%
`22%
`Bone only
`l 3 %
`16%
`21 %
`Visceral only
`38%
`39%
`44%
`Mixed
`
`
`
`2% 2%Not evaluable 0
`
`49%
`27
`51%
`36%
`30%
`60%
`
`62%
`42%
`34%
`81%
`19%
`
`42%
`59%
`54%
`21 %
`
`39%
`28
`61%
`51%
`28%
`6 1%
`
`54%
`37%
`39%
`75%
`25%
`
`42%
`56%
`43%
`18%
`
`- 42%
`32
`58%
`37%
`26%
`64%
`
`58%
`41%
`38%
`79%
`21%
`
`42%
`62%
`42%
`19%
`
`‘For patients relapsing on or after adjuvant tamoxifen. 1 Complete and partier response in patients treated for
`advanced disease. PS, performance status; ER, oestrogen receptor; PR, progesterone receptor.
`
`treatment groups (Table 2). Since partial response was not
`allowed for patients with evaluable non-measurable lesions,
`almost all responders in this subgroup were assigned a stable
`disease category. This included patients with only bone osteo—
`lytic lesion. Response rates including patients with stable
`diseaSe for >6 months were 34.1% for anastrozole 1 mg,
`33.9% for anastrozole 10 mg and 32.8% for megestrol acetate.
`There was no significant difiercnce among the three treatment
`arms in their response rate in subgroups of patients according
`to the presence or absence of measurable lesions, disease
`status, receptor status and prior response to hormonal therapy
`(Table 3). Responses were observed in patients progressing
`on adiuvant tamoxifen as well as in patients who received
`tamoxifen treatment 'for advanced 'disease. The highest
`response rate” was achieved in the subgroup of patients with
`soft tissue only disease (Table 3). Since the protocol did not
`specify a specific tamoxifen'withdrawal interval, we analysed
`the response rate according to whether patients had short (<3
`months) or long (>3 months) tamoxifen withdrawal interval
`before entering the study and there was no difference
`
`The median duration of response was 261 days for anastro—
`zolc 1 mg, 257 days for mcgestrol acetate and was not reached
`at the time of the analysis for anastrozole 10 mg. The duration
`of response was greater than 24 weeks in 74% of patients
`responding to anastrozole treatment. There was a high rate of
`completion of quality of life questionnaires throughout the
`, study. The percentage of patients who completed the ques-
`tionnaire from the total number of expected patients at each.
`follow-up timcpoint was more than 90% at entry, more than
`75% at week 12 and more than-70% at week 24. There were
`no differences between the treatment groups in the physical or
`the functional dimensions of the quality of life questionnaire.
`At week 12,
`there was statistical evidence that megestrol
`acetate was associated with some benefit in the psychological
`dimension compared with anastrozolc at 1 mg (P: 0.008) or
`10 mg (P':0.003Z).' However, this difference was not apparent
`at 24 weeks. Subjective symptom scores revealed no difference
`between treatments in analgesic use. Anastrozole 10 mg was
`associated with less bone pain at 12 Weeks than megestrol
`acetate (1’: 0.011). Anasttozole 1 mg was associated with
`
`AstraZeneca Exhibit 2138 p. 4
`
`

`

`, 408.
`
`_
`
`W. Jonat er a1.
`
`acetate; headache and hot flushes on anastrozole 10 mg (5%
`each); and nausea on anastrozole 1 mg (4.5%). The incidence
`of side-effects was, therefore, low for both anastrozole doses.
`Headaches, hot flushes and nausea were all described as mild
`or moderate and transient in nature. With the exception of
`more weight gain and oedema in the megestrol acetate group
`compared to anastrozole 1 mg, there were no significant differ-
`ences between the treatment groups in side-effects (Table 4).
`The numbers of patients with absolute weight gain of at least
`5 or 10% from baseline were also statistically significantly
`greater on megestrol acetate compared with either close of
`anastrozole (Figure 23). In addition, weight continued to be
`gained with time whilst patients were being treated with
`mcgestrol acetate (Figure 2b), In general, at the time of
`this analysis, all three treatments were well tolerated. The
`incidence of withdrawals because of adverse events irrespec-
`tive of causality was low: 3% for anastrozole 1 mg (4 patients),
`3.4% for anasnrozole 10 mg (4 patients) and 4.8% for meges-
`trol acetate (6 patients), with no particular adverse event
`predominating in any group.
`
`DISCUSSION
`
`.
`
`Aminoglutethirnide which represented the first generation of
`aromatase inhibitors has demonstrated clinical activity at least -
`similar to tamoxifen and progestins in several randomised .
`trials [2, 5, 21], but it has a poor side—eflect profile, and is
`often prescribed with corticosteroid replacement due to its
`non-selectivity [2, 8, 9]. More recently, a selective second
`generation arozmatase inhibitor (4-hydroxyandroste‘nedione)
`has been clinically investigated in breast cancer, mainly in
`Europe, and it induced clinical remissions in uncontrolled
`trials that were comparable with published data with other
`hormonal agents [22, 23]. One controlled study suggested
`comparable eflicacy to tamoxifen [24]. However, 4-hydroxy—
`androstenedionc has a poor oral bioavailability, limiting its
`use to parenteral treatment which is associated with local
`injection reactions [22: 23]. A number of new third generation
`aromatase inhibitors which are potent, selective, and orally
`bioavailable are currently under clinical investigation [25].
`This study reports efficaag of the new selective aromatase
`inhibitor, anasrrazole compared with a standard hormonal
`treatment, megestrol acetate. Time to treatment progression,
`time to treatment failure, survival and objective response rate
`were similar for the three treatment groups.
`Outcome of advanced breast cancer can be related more to
`patient characteristics than to treatment differences. In this
`
` H Anaslrozole 1mg
`
`—‘‘‘‘‘‘ Anasrrozole 10mg
`- — - Megesuol acetate
`
`0.1
`
`0
`
`
`—1 —I— r
`l
`r
`I
`'1'— r—— T—r— ‘ r
`50
`100
`150
`200
`350 300
`350 400
`450
`500 550
`Time to progression (days)
`
`
`2-!)
`
`E 2E E
`
`:
`
`I)
`
`
`
`Proportionalive
`
`
`
`0.3
`
`-— Anaslrozole 1mg
`w Anasrrozole 10mg
`- - - Megestrol acetate
`
`0
`
`' fi—l—F_T—l_—l—lfi——‘r—T‘~—r
`50
`100
`150
`200 250 300 350 400
`451) 500 550
`Time [0 death (days)
`
`Figure l. (a) Kaplan—Meier probability oftime to progression.
`(1)) Kaplan—Mcier probability oftime to death.
`
`better performance status scores at 12 weeks than megestrol
`acetate (P: 0.007) and the odds ratio still favoured anastro—
`zole 1 mg at 24 weeks, although the diflerenee did not reach
`the critical
`level of statistical significance for the analysis
`(P= 0.046). Both anastrozole doses produced consistent sup-
`pression of oestradiol levels to below the limit of detection of
`the assay in more than 90% of patients during the treatment
`period.
`
`Tolerabilizy
`The most frequently reported adverse events which were
`considered by the investigators to be probably drug-related
`were weight gain (8%) and dyspnoea (5.6%) on megestrol
`
`—————__—_—___
`Table 2. Best objecn'oe responrefar all randomisedpatients
`
`Megestrol acetate
`Anastrozole
`Anastrozole
`1 mglday
`10 mg/day
`40 mg four times
`
`(n: 135)
`(n = 113)
`daily (It: 125)
`
`46 (34.1)
`Response rate (CR+PR+ SD >6 months) (96)
`2 (1.5)
`Complete response (%)
`-
`12 (8.9)
`Partial response" (%)
`32 (23.7)
`Stable disease 26 months (%)
`10 (7.4)
`Stable disease <6 months (%)
`—~——‘———————————_—
`Progression (7°)
`79 (58.5)
`
`41 (32.8) ‘
`40 (33.9)
`' 3 (2.4)
`3 (2.5)
`10 (8)
`12 (10.2)
`28 (22.4)
`25 (21.2)
`14 (11.2)
`18 (15.3)
`
`60 (50.8)
`70 (56)
`
`‘Partial response category was not allowed for any patient with evaluable non—measurable lesions only (includes
`patients with osteolytic bone lesions).
`
`AstraZeneca Exhibit 2138 p. 5
`
`

`

`Arimidex Versus Megestrol Acetate 111 Breast Cancer
`
`409
`
`Table 3. Objective response rate (complete plus partial response) in dyfi‘erent patient subgroups
`
`Anastrozole
`Anastrozole
`
`
`
`
` Subgroup 1 mg/day 10 mg/day
`
`Mcgestrol acetate
`40 mg four times daily
`
`Receptor status
`ER+
`ER unknown/ER—‘
`Disease status
`Adiuvant tamoxifen only
`Tamoxifen for advanced disease
`Prior response to tamoxifen
`Prior response
`No prior response
`Sites of disease
`Soft tissue only-
`Bone only-f
`Visceral only
`Mixed
`Patients with measurable disease
`Patients with measurable non-visceral
`lesions
`
`9/84 (11%)
`5/51 (10%)
`
`8/66 (12%)
`6/69 (9%)
`
`4/25 (16%)
`2/44 (5%)
`
`7/15 (47%)
`1/30 (3%)
`2/28 (7%)
`4/60 (7%)
`13/109 (12%)
`9/45 (20%)
`
`7/64 (11%)
`8/53 (15%)
`
`4/45 (9%)
`11/72 (15%)
`
`6/37 (16%)
`5/35 (14%)
`
`8/22 (36%)
`1/29 (3%)
`0/19
`6/46 (13%)
`14/89 (16%]
`10/46 (22%)
`
`9/72 (13%)
`4/53 (8%)
`
`6/52 (12%)
`7/73 (10%)
`
`3/27 (11%)
`4/46 (9%)
`
`6/25.(24%)
`2/36 (6%)
`1/16 (6%)
`4/48 (8%)
`11/99 (11%)
`7/50 (14%)
`
`‘Padcnts with known ER negative status were/entered into the study only if they had evidence of prior response
`to tamoxifen. fPatients with bone lesions were evaluable but non-measurable (PR was not allowed).
`
`Table 4. Prospective analysis afphannacological adverse event: (irrespective ofcausality) .
`
`Statistical analysis
`(P value”)
`Percentage and incidence
`Anastrozole
`Anastrozole
`Megestrol
`1 mg versus
`10 mg versus
`acetate
`40 mg four times
`megestrol
`megestrol
`
` daily (11 = 125) acetate acetate
`
`
`
`Anastrozole
`10 mg/day
`(n = 117)
`
`Anastrozole
`1 mg/day
`(n = 134)
`
`Anticipated effects
`P:O.170
`P=0.045
`8
`3.4
`2.2
`Weight gain
`P=0.839
`11:0,022
`10.4
`12.0
`3.0
`Oederna
`P=().121
`P=O.763
`4.8
`0.9
`3.7
`'l‘hrombo—cmbohc disease
`H.252
`H.269
`16.0
`22.2
`21.6
`Gastrointestinal disturbance*
`P=0.786
`[H1242
`6.4
`5.1
`3.0
`Hot flushes
`
`Vaginal dryness 11—1000 0.7 0.9 0.8 P=1.000
`
`
`
`
`
`*Includcs a grouping of all GI disturbances such as nausea, vomiting, anorexia, diarrhoea, constipation and dyspepsia.
`
`there were no clinically significant differences in the
`trial,
`major prognostic factors among the three randomised groups, .'
`and all objective eflicicacy analyses were conducted on an
`intention to treat basis and were adjusted for receptor status
`and disease status. It is well known that interpretation of
`response criteria can vary considerably among clinicians ’and'
`that patient selection can significantly influence the response
`rate in clinical'trials [26]. In this open vmulticentre trial,
`variation of interpretation of response criteria was avoided
`by assigning response only through a computer programme,
`which strictly assessed response from percentage tumour -
`regression of verified measurements for measurable lesions. .
`Two successive measurements of at least 4 weeks apart con-
`firming response were needed for the patient to qualify as a
`responder. Standard UICC criteria were used for assessment
`of response in measurable lesions [19] ,~ but, we did not allow
`the category ofpartial response to be assigned to patients with
`
`only evaluable but non-measurable lesions as it is difficult to
`ensure the obiectivity of partial response in these patients.
`Since those patients represented approximately 20% or more
`of the different treatment groups, this resulted in lower num—
`bers of assigned partial responses. Most of the- responding
`patients with evaluable non-measurable disease were counted
`as stable disease. All patients who entered the trial were in
`progression after prior tamoxifen, and stabilisation of disease
`for a long period confers a clinical benefit thathas been shown '
`, to be equal to that of partial response [27, 28]. Therefore, we
`considered the overall response rate of complete response,
`partial response and stable disease (for 26 months) in pre- .
`senting the data. Although the complete and partial response
`rate appears low, it is in accordance with published response
`rates for hormonal agents given as second—line therapy after
`failure of tamoxifen in an unselected patient population. Such
`trials reported a 543% response rate for-megestrol acetate
`
`AstraZeneca Exhibit 2138 p. 6
`
`

`

`.
`
`advanced malignancy or other terminal illness such as AIDS
`[34, 35]. In this trial, as is typically the case with breast cancer
`patients who are candidates for hormonal therapy, the patient
`population was not cachectic as they had good performance
`status and were of expected or higher mean weight for their
`height (Table 1). There is some literature evidence fi'om a
`small number of patients on megestrol acetate at doses of
`800 nag/day or higher that weight gain is primarily the result
`of an increase in body mass and not iust water retention [3 6].
`However, in our trial, weight gain was also associated with
`oedema in many cases (Table 4). Patients continued to gain
`weight on treatment with megestrol acetate-(Figure 2b), which
`represents a disadvantage for responding patients who con-
`tinue treatment: for longer periods. This pattern of continued
`weight gain with prolonged exposure to progestins has already
`been reported [4].
`The most frequent adverse events reported as related to
`anastrozole by the investigators were headaches and 'hot
`flushes at 10 mg (5% each) and nausea at 1 mg (4.5%). These
`events may relate to the change in hormonal status and the
`result of oestrogen withdrawal as apharmacological effect of
`anastrozole. Surprisingly, the incidence of classical oestrogen
`withdrawal symptoms, such as hot flushes and vaginal dryness,
`was not statistically higher for either anastrozole dose than for
`megestrol acetate. Gastro—intestinal adverse events, irrespec-
`tive of causality, were slightly higher for anastrozole patients,
`but the difference was not statistically significant and the
`~ events mOStly comprised of mild transient nausea. In addition
`to the primary endpoints of efieacy and tolerabiliry, this trial
`also looked at theendpoints of survival and quality of life. At
`the time of analysis, the percentage of deaths was 15.6% for
`anastrozole 1 mg, 18.6% for anastrozole 10 mg and 22.4% for »
`megestrol acetate. The lack of a difference in other objective
`eflicacy endpoints between the three treatment groups in this
`trial makes it unlikely that a survival difference will emerge
`even with more mature survival data.
`Quality of life was assessed by the RSCL and subjective
`symptom scores. There was statistical evidence of a significant
`advantage in the psychological dimension for megestrol acet—
`ate at 12 weeks which was not apparent at 24 weeks. Alternar
`tively, there was statistical evidence of a significant advantage
`for anastrozole 1 mg in terms of performance status and for
`anastrozole 10 mg for improvement in bone pain, both at 12
`weeks. Analysis at 24 weeks of these parameters still favoured
`anastrozole, but did not reach statistical-significance. The
`clinical significance of these findings is uncertain since most
`