Dose-Response Trial of Megestrol Acetate in Advanced
`Breast Cancer: Cancer and Leukemia Group B
`Phase III Study 8741
`
`By Jeffrey Abrams, Joseph Aisner, Constance Cirrincione, Donald A. Berry, Hyman B. Muss M. Robert Cooper,
`|. Craig Henderson, Lawrenoe Panasci, Jeffrey Nrshner, John Ellerton, and Larry Norton
`
`To investigate whether dose escalation of
`Purpose:
`megestrol acetate (MA) improves response rate and
`survival in comparison with standard dosesof MA.
`PatientsandMethods: Threehundredsixty-eightpa-
`tlents with metastatic breast cancer, positive and/or
`unknow n estrogen and progesterone receptors, zero or
`one priortrialofhormonaltherapy,and no prior chemo-
`therapy for metastatic disease were prospectively ran-
`domized into three groups. The groups of patients re-
`ceived either MA 160 mg/d (one tablet per day), MA
`800 mg/ d (five tablets per day), or MA 1,600 mg/d (10
`tabletsper day).
`Results: Patient characteristics were well balanced
`in the three treatment groups. Three hundred sixty-six
`patients received treatment and were included in the
`analyses. The response rateswere 23%, 27%, and 27%
`for the 160-mg, 800-mg, and 1,600-mg arms, respec-
`tively. Response duration correlated inversely with dose.
`
`Median durations of response were 17 months, 14
`months, and 8 months for the 160-mg, 800-mg, and
`1,600-mg arms, respectively. No significant differences
`in the treatment arms were noted for time to disease
`
`progression or for survival; survival medians were 28
`months(low dose), 24 months(mid dose)and 29 months
`(high dose). The most frequent and troublesome toxic-
`ity, weight gain, wasdose-related, with approximately
`20% of patients on the two higher—dose arms reporting
`w eight gain of more than 20% oftheir prestudy weight,
`compared w ith only 2% in the 160-mg dose arm.
`Conclusion: With a median follow-up of 8 years,
`these results demonstrate no advantage for dose esca-
`lation of MA in the treatment of metastatic breast can-
`cer.
`
`J Clin Oncol17:64-73. ©1999 by American Society of
`Clinical Oncology.
`
`EGESTROL ACETATE (MA), a semisynthetic, oral
`progestin with excellent gastrointestinal absorption,
`has demonstrated efficacy comparable to that of tamoxifen
`as first-line treatment
`in women with advanced breast
`
`cancer.1 At the standard dose of 160 mg daily, side effects
`from MA are minimal and include nausea, vaginal bleeding
`and discharge, weight gain, and fluid retention. In a series of
`908 patients treated at this dose, the only serious toxicities
`possibly attributable to MA were three cases of congestive
`heart failure, one case of puhnonary embolism, and one case
`of acute left-ventricular failure.2
`
`Efforts to improve the activity of MA focused on dose
`escalation, an approach that has been widely pursued for
`cytotoxic agents but less so for hormonal therapies. Potential
`
`
`From the University of Maryland Cancer Center, Baltimore, MD;
`Cancer and Leukemia Group B Statistical Office, Durham, NC;
`Bowman Gray School of Medicine, Wnston-&lem, NC; Dana-Farber
`Cancer Institute, Boston, MA; McGill Cancer Center, Montreal, Canada;
`State University of New York at Syracuse, Syracuse, NY; University of
`California at San Diego, San Diego, CA; and the Memorial Sloan-
`Kettering Canoer Center, New York, NY.
`Submitted April 10, 1998; accepted $ptember 1, 1998.
`Address reprint requests to Jeffrey Abrarrs MD, National Cancer
`Institute, 6130 Executive Blvd, EPN 741, Rookville, MD 20892—7436;
`Email AbramsJ@CTEP.nci.nih.gov.
`to 1999 by American Sbciety of Clinical Oncology.
`0732-183X/99/1 701 -0064$3.00/0
`
`mechanisms that could explain improved antitumor effects
`from high-dose progestins include an enhancement of the
`direct cytotoxic effect of the honnone via the progestin
`receptor,3 additional direct cytotoxic effects on other related
`and important corticosteroid receptors,4 and an increase in
`the indirect effects of progestins by further suppression of
`the hypothalamo-pituitary-adrenal axis.5 Two phase III trials
`compared high versus standard doses of another progestin,
`medroxyprogesterone acetate, administered by intramuscu-
`lar injection, and demonstrated improved response rates.“
`These encouraging results provided the impetus to initiate
`further trials to explore the dose-response effect of MA.
`However, not all investigators found higher doses to be more
`effective.8=9
`
`The superior oral availability of MA compared with
`medroxyprogesterone acetate led the Piedmont Oncology
`Association (POA) to perform a phase III trial in patients
`with metastatic breast cancer comparing high-dose (800
`mg/d) with standard-dose (160 mg/d) MA. All but two of the
`172 patients entered had one prior trial of tamoxifen therapy
`for either metastatic (74%) or adjuvant (26%) treatment.
`High-dose MA resulted in a superior complete plus partial
`response rate (27% V 10%, P : .005), time to treatment
`failure (median, 8.0 V 3.2 months, P :.019), and survival
`(median, 22.4 V 16.5 months, P : .04) when compared with
`standard-do se therapy. 10
`
`64
`
`Journal of Clinical Oncology, Vol 17, No 1 (January), 1999: pp 64-73
`
`Downloaded from ascopubs.org by 151.194.33.114 on March 28, 2017 from 151.194.033.114
`Copyright © 2017 American Society of Clinical Oncology. All rights reserved.
`
`AstraZeneca Exhibit 2150 p. 1
`InnoPharma Licensing LLC v. AstraZeneca AB IPR2017-00900
`Fresenius-Kabi USA LLC v. AstraZeneca AB IPR2017-01913
`
`

`

`MEGESTROL ACEIATE IN MEIAS'IATIC BREAST CANOE?
`
`65
`
`A phase l/H trial demonstrated that the dose of MA could
`be further escalated to 1,600 mg daily.“>12 Although dose-
`limiting toxicity was not reached, substantial weight gain
`(median, 5.0 kg; range, 5.6 to 44 kg) occurred in 71% of
`patients at the 1,600-mg dose level, making further escala-
`tion difficult. Responses occurred at all dose levels in this
`trial, but the most promising results occurred in a subset of
`27 patients who had progressed after treatment with standard
`doses ofMA (160 mg/d). A 15% response rate (one complete
`response, three partial responses) was noted in this subset,
`and 10 patients (37%) had stable disease lasting a median of
`5.4 months. Two of the objective responses occurred in
`women whose tumors had not previously responded to the
`standard MA dose.
`
`These provocative results provided the rationale for the
`Cancer and Leukemia Group B (CALGB) to develop a
`randomized,
`three-arm, phase III
`trial design in which
`women with metastatic breast cancer received either standard-
`
`dose MA (160 mg/d), five times the standard dose (800
`mg/d), or 10 times the standard dose (1,600 mg/d). A
`preliminary report of this trial has been presented.13
`
`FATIENTS AN D MEI'HODS
`
`Patient Selection
`
`Women at least 18 years of age and with histologically documented
`breast carcinoma and progressive metastatic disease were eligible.
`Other requirements included a performance status of 0 to 3, positive
`and/or unknown estrogen and progesterone receptors, and no prior or
`concomitant malignancy other than curatively treated in situ cancer of
`the cervix or basal cell carcinoma of the skin. Patients with metastatic
`disease that was either bidimensionally measurable or assessable were
`entered, but they were assessed separately. Only one prior hormonal
`therapy (drug or surgical manipulation) for adjuvant or metastatic
`treatment was permitted, excepting progestins, which were not allowed.
`Patients who responded to their initial hormonal therapy were required
`to wait 6 weeks ofi all therapy before entering the study to avoid
`confusion with a withdrawal response. This did not apply to surgical
`manipulations. Prior adjuvant chemotherapy was acceptable if the
`disease-free period of treatment was more than 1 year, but chemother-
`apy for metastatic disease was not permitted.
`Brain, leptomeningeal, lymphangitic, lung, and extensive liver and
`bone marrow metastases were all causes for exclusion. Evidence of
`
`liver, and bone marrow function was required, and
`normal kidney,
`patients with serious medical problems (especially congestive heart
`failure, uncontrolled hypertension, or diabetes mellitus) and those with
`ahistory of thrombophlebitis or stroke were excluded. All participants
`had to sign an informed consent that had been approved by institutional
`review boards.
`
`weeks for responding patients who were 011 treatment for long periods
`of time.
`
`Treatment and Dose Modification
`
`Patients were stratified according to prior hormone therapy (yes V
`no), prior adjuvant therapy (yes V 110), and receptor status (estrogen-
`and/or progesterone-positive V both unknown). They were then random-
`ized within each stratification subgroup to one of the following three
`treatment arms: (1) MA at 160 mg/d, one tablet per day; (2) MA at 800
`mg/d, five tablets per day; and (3) MA at 1,600 mg/d, 10 tablets per day.
`To facilitate oral administration, Bristol-Myers Squibb Co (Princeton,
`NJ) provided specially prepared 160-mg tablets for this trial. A prior
`study in normal volunteers compared the phannacokinetic profile of the
`40-mg four-times-a-day dosage with the 160-mg investigational tablet.
`It demonstrated that the peak plasma concentrations,
`the extent of
`absorption (area under the curve), and the half-life (t 1/2) showed no
`significant differences. The bioequivalence 0f the 160-mg investiga-
`tional tablet was 97% of the 40-mg four-times-a-day dosage.14 The
`recommended administration schedule was as follows: arm 1, one tablet
`at 8 AM; arm 2, one tablet at 8 AM, 12 PM, and 6 PM, and two tablets at 10
`PM; and arm 3, two tablets at 8 AM and 12 PM and three tablets at 6 PM
`and 10 PM. No dose modifications were allowed for toxicity. Uncon-
`trolled arterial hypertension, congestive heart failure, and thrombophle-
`bitis were all causes for discontinuation of MA. Patients with weight
`gain and minor vaginal bleeding were encouraged to continue their
`treatment. Ifunacceptable side effects occurred in a responding patient,
`the patient was taken off the study drug and could then be treated with
`additional hormonal agents at the physician’s discretion. After disease
`progression, patients were to be removed from MA and treated at the
`physician’s discretion. For patients with rapidly progressive or life-
`threatenirrg disease, chemotherapy was usually initiated, whereas
`further hormonal manipulations were considered for those with slower
`progression.
`
`Study Design
`
`The objective of this trial was to compare the response rates of the
`three treatment arms. The response rates were hypothesized to fall
`between 25% and 55%. A sample size of 100 patients per arm would
`yield at least an 80% power to detect a 17% additive difierence, such as
`30% versus 47%, between any two arms. Thus, with 123 patients
`randomized to the 160-mg arm, 124 to the 800-mg arm, and 119 to the
`1,600-mg arm, there is better than 80% probability of detecting a 17%
`difierence in response between any two arms.
`
`Variabl as of Interest
`
`We examined several demographic, pretreatment clinical, and treat-
`ment-related variables. These included: patient age, menopausal status,
`receptor status, performance status, measurability of disease. sites of
`involvement, number of involved sites, prior treatment, and time from
`initial diagnosis of breast cancer to first recurrence.
`
`End Points
`
`Treatment Eval uati on
`
`Before beginning treatment with MA, all patients had a chest x-ray,
`electrocardiogram, bone scan, and a liver/spleen scan or abdominal
`computed tomogram. Additional studies were performed as necessary
`to measure indicator lesions. Patients were initially evaluated for
`response and toxicity every 4 weeks, but this was changed to every 8
`
`The study end points were response rate, response duration, time to
`disease progression, and overall survival. Definitions of response were
`assessed according to standard response criteria for patients with
`bidimensionally measurable disease.15 Briefly, a complete response was
`defined as the complete disappearance of all signs and symptoms of
`disease, including reossification of osteolytic lesions, for at least 30
`days. Apartial response was defined as a decrease of at least 50% in the
`
`Downloaded from ascopubscrg by 151.194.33.114 on March 28, 2017 from 151.194.033.114
`Copyright © 2017 American Society of Clinical Oncology. All rights reserved.
`
`AstraZeneca Exhibit 2150 p. 2
`
`

`

`66
`
`ABRAMS EI'AL
`
`product of the cross-perpendicular dimensions of all measurable
`indicator lesions, and no worsening of any existing lesion or the
`appearance of any new lesions, for at least 30 days. Stable was defined
`as less than a 50% reduction or less than a 25% increase in the product
`ofthe cross-perpendicular dimensions of all measurable lesions, with no
`new lesions for at least 60 days. For patients with assessable disease,
`symptomatic benefit but continued presence of lesions without the
`appearance of new lesions for at least 60 days qualified as stable
`disease. Progression was defined as an increase of at least 25% in the
`product of the cross-perpendicular dimensions of any measured lesion
`over the size present at entry, or for responders, the size at the time of
`maximum regression or the appearance of any new lesion. Those with
`only assessable disease were excluded from the assessment of partial
`responses.
`Duration of MAresponse was measured from the date of complete or
`partial response, whichever occurred first, on MA until disease progres-
`sion. Responders who were alive and disease-free were censored at their
`last follow-up visit. Time to disease progression was the date of study
`entry until the date of first disease prOgression or the date of last
`follow-up for patients who were alive and disease-free. Overall survival
`was the date of study entry until death due to any cause, or until date of
`last follow-up for survivors.
`
`Statistical Methods
`
`Survival curves were estimated by the Kaplan-Meier product limit
`method, whereas the log-rank test was used to compare two or more
`survival distributions. We used the Cox proportional hazards model
`initially to screen for individual variables potentially related to survival
`and time to disease progression. Subsequently. this method was used to
`identify sets of prognostic variables while controlling for the effect of
`other variables in the model.”
`Comparisons between categorical Variables were performed bythe X2
`test. The Mantel-Haenszel X2 test measures whether tumor response in-
`creases with increased MA dose (treatment arm). All P values are two-sided.
`We define statistical significance as P S .05. We included data available
`as of February 1997. The median follow-up was 8.2 years.
`
`Comparison with POA Trial
`The POA kindly provided us with data from their previously
`published study.10 We included in analysis only variables that both our
`study and the POA study measured. We defined variables identically in
`the two studies, both pretreatment characteristics and end point measures.
`
`Patient Characteristics
`
`RESULTS
`
`A total of 368 patients were enrolled from 23 CALGB
`main member institutions and their affiliates between June 1,
`1987, and March 22, 1991. Of these, 124 patients were
`randomly assigned with equal probability to the l60-mg
`arm, 124 were assigned to the SOO-mg arm, and 120 were
`assigned to the 1,600-mg arm. Two patients, one on the
`160-mg arm and the other on the 1,600-mg arm, never
`received treatment. Twenty-eight patients did not meet the
`eligibility criteria. For the l60—mg arm (n : 7), reasons for
`ineligibility included the following:
`two prior hormone
`regimens (three patients),
`laboratory values out of range
`(one patient), less than 1 year from end of adjuvant treatment
`
`(one patient), visceral crisis (one patient), and prior chemo-
`therapy for metastatic disease (one patient). For the BOO-mg
`arm (n : 9), reasons for ineligibility included the following:
`two prior hormone regimens (3 patients), laboratory values
`out of range (one patient), less than 1 year from end of
`adjuvant treatment (three patients), and Visceral crisis (two
`patients). For the 1,600-mg ami (n : 12), reasons for
`ineligibility included the following: less than 1 year from
`end of adjuvant
`treatment (two patients), visceral crisis
`(seven patients), prior chemotherapy for metastatic disease
`(one patient), estrogen— and progesterone-negative (one
`patient), and prior cancer (one patient).
`In keeping with an intent-to-treat analysis, all treated
`patients (n : 366) were included in analyses. One patient
`received radiotherapy (RT) concurrent with MA therapy.
`She is excluded from analyses of MA response but included
`in analyses of other end points. Analyses (not shown) of all
`major study end points were recalculated using only eligible
`patients. No significant differences were found between
`these results and those based on the intent-to—treat principle.
`Table 1
`lists the pretreatment characteristics of partici-
`pants by treatment arm. Patients on the three arms were well
`matched at pretreatment. Examination of baseline character-
`istics between white and nonwhite patients revealed no
`significant differences. The relatively small size of the
`nonwhite populations did not allow us to perform a separate
`analysis of outcome according to race.
`
`Tumor Response and Duration of Response
`
`Table 2 lists the maximum tumor response to MA therapy.
`Of the 365 total patients, 19 were missing follow-up tumor
`assessments. The most common reason for missing assess-
`ments was treatment tennination before the scheduled 8-week
`
`follow-up evaluation. Reasons for early treatment termina-
`tion included early death, toxicity, withdrawn consent. physi-
`cian decision, and other complicating illness. Patients with-
`out repeat tumor assessments due to early death or MA toxicity
`were considered nonresponders and were included in re-
`sponse rate calculations. Other patients were considered to
`have nonassessable tumors and were excluded from response
`rates. Thus, response evaluation is based on 357 patients.
`Ninety-one patients achieved a tumor response, that is,
`either a complete or partial response, while on MA. The
`response rates were 23%, 27%, and 27% for the l60-mg,
`800-mg, and 1,600—mg arms, respectively. Results ofthe Mantel-
`Haenszel test show that response rate did not significantly
`rise with increasing doses of MA. The overlapping 95%
`confidence intervals indicate that treatment did not correlate
`
`with tumor response. Note that including nonassessable
`patients in response rates gave similar results as did exclud-
`ing them.
`
`Downloaded from ascopubsprg by 151.194.33.114 on March 28, 2017 from 151.194.033.114
`Copyright © 2017 American Society of Clinical Oncology. All rights reserved.
`
`AstraZeneca Exhibit 2150 p. 3
`
`

`

`MEGESTROL ACEIATE IN MEIAS'IATIC BREAST CANOE?
`
`67
`
` 'lable 1. Patient Demographics and Pretreatment Giaracteristics 'lable 1. Patient Demographics and Pretreatment OiaraderistirsCont’d)
`
`
`Megestrol Acetate Treatment Arm
`Megestrol Acetate Treatment Arm
`
`160 mg
`800 mg
`1,600 mg
`Total
`160 mg
`800 mg
`1,600 mg
`Total
`No.
`%
`No.
`%
`No.
`%
`No.
`%
`Variable
`No.
`%
`No.
`%
`No.
`%
`No.
`%
`Variable
`
`
`'lbtal patients
`Age, years
`< 40
`40-49
`50-59
`60-60
`70+
`Median
`finge
`Menopausal status
`Pre-
`Reri-l poS-
`Unknown
`Estrogen receptor satus
`Negative
`Fbs'tive, borderline
`Unknown
`Progesterone receptor
`status
`
`_
`Negétwe
`Fbstive, borderline
`Unknown
`CALGB performance
`more
`
`123
`
`100
`
`124
`
`100
`
`119
`
`100
`
`366
`
`100
`
`5
`9
`25
`53
`31
`
`4
`7
`20
`43
`25
`
`65
`35-88
`
`2
`19
`26
`42
`35
`
`2
`15
`21
`34
`28
`
`63
`37-90
`
`1
`12
`32
`37
`37
`
`65
`38-89
`
`1
`10
`27
`31
`31
`
`7
`92
`1
`
`5
`76
`19
`
`9
`114
`0
`
`4
`95
`24
`
`18
`67
`38
`
`7
`93
`0
`
`3
`77
`20
`
`15
`54
`31
`
`12
`111
`1
`
`10
`90
`< 1
`
`8
`110
`1
`
`7
`88
`29
`
`16
`74
`34
`
`6
`71
`23
`
`13
`60
`27
`
`6
`9O
`23
`
`20
`68
`31
`
`8
`42
`82
`131
`103
`
`2
`11
`22
`36
`28
`
`64
`35-90
`
`8
`29
`92
`335
`2 < 1
`
`17
`273
`76
`
`4
`75
`21
`
`15
`57
`28
`
`55
`37
`5
`
`54
`17
`57 209
`26
`103
`
`Prior therapy
`Prior chemotherapy
`No
`Yes
`Unknown
`Prior radiation
`therapy
`No
`Yes
`Unknown
`Prior hormone
`therapy
`No
`Yes
`Unknown
`Prior tamoxifenT
`3
`7
`4
`3
`3
`2
`3
`2
`No
`Yes
`73
`97
`72
`96
`71
`96
`216
`96
`Unknown
`0
`0
`1
`1
`0
`0
`1
`1
`
`
`61
`39
`0
`
`47
`53
`0
`
`40
`60
`0
`
`73
`45
`1
`
`65
`54
`0
`
`44
`74
`1
`
`61
`38
`1
`
`55
`45
`0
`
`61
`223
`39
`141
`2 < 1
`
`52
`189
`48
`176
`1 < 1
`
`37
`63
`< 1
`
`39
`141
`61
`224
`1 < 1
`
`74
`48
`1
`
`66
`56
`1
`
`48
`75
`0
`
`60
`39
`1
`
`54
`46
`< 1
`
`39
`61
`0
`
`76
`48
`0
`
`58
`66
`0
`
`49
`75
`0
`
`liver, brain, adrenal,
`*Viweral metastases lung, pleura, pleural effusion,
`.
`.
`.
`.
`.
`.
`kidney, spleen, pancreas brachlal, malignant axites pericardium, mesentary,
`.
`.
`hilar ma§1 Bone metastases bone, bone marrow. Soft “$.19 metastases
`.
`.
`.
`recurrent primary, inoperable primary, oontralateral breast, chefi wall, nodes,
`axilla, skin, scalp, dwoulder.
`TArnon
`atientswho received rior hormonal thera
`p
`
`g p
`
`M”
`
`0
`1
`2
`
`3
`Unknown
`Di$ase type
`A” aWb'e
`Any ma‘rab'e
`Disease-free interval
`(diagnosisto 151
`recurrence)
`
`0
`0-2 Years
`2 2 years
`Metaéatic involvement*
`Viweral metastases
`No
`
`66
`48
`4
`
`4
`1
`
`50
`73
`
`13
`27
`83
`
`54
`39
`3
`
`3
`1
`
`41
`59
`
`11
`22
`67
`
`65
`45
`10
`
`3
`1
`
`47
`77
`
`15
`47
`62
`
`53
`36
`8
`
`2
`1
`
`3'3
`62
`
`12
`38
`50
`
`70
`41
`5
`
`2
`1
`
`43
`76
`
`9
`25
`84
`
`59
`34
`4
`
`2
`1
`
`35
`64
`
`8
`22
`71
`
`201
`134
`1 9
`
`9
`3
`
`140
`226
`
`37
`100
`229
`
`2
`1
`
`38
`62
`
`10
`27
`63
`
`76
`
`62
`
`72.
`
`5,3
`
`68
`
`57
`
`216
`
`59
`
`Yfi
`Bone metastases
`No
`Ya
`Soft tissue metasiasg
`No
`Y‘s
`'t
`N .
`fmet s t'
`o 0 study :nlfys' $81
`1
`2
`3+
`2
`2
`2
`Media“
`
`finge 1-5 1-5 1-5 1-5
`
`
`
`
`47
`43
`80
`
`69
`54
`
`14
`7o
`39
`
`38
`35
`65
`
`56
`44
`
`11
`57
`32
`
`t
`
`52
`46
`78
`
`65
`59
`
`13
`69
`42
`
`42
`37
`63
`
`52
`48
`
`10
`56
`34
`
`51
`48
`71
`
`65
`54
`
`11
`74
`34
`
`43
`40
`60
`
`55
`45
`
`9
`62
`29
`
`15°
`137
`229
`
`199
`167
`
`38
`213
`115
`
`41
`37
`63
`
`54
`46
`
`10
`53
`31
`
`_
`_
`Figure 1 shows the duration of response on MA for the 91
`patients whose tumors responded to MA treatment. Dose did
`not correlate with achieving a response; however, dose did
`correlate with length of time in response. Of interest, the
`.
`.
`.
`.
`correlation was negative (P < .003),
`that is, the higher
`the MA dose, the shorter the time in response. Specifically,
`the median response duration was 17 months for patients on
`the l60-mg ami, 14 months for patients on the 800-nig arm,
`and 8 months for patients on the 1,600-mg arm.
`.
`.
`.
`Timeto Disease Progress on
`
`Figure 2 shows time to disease progression by treatment
`7
`.
`.
`_
`.
`1
`.
`arm. Two patients from the mid-dose arm discontinued ,V A
`therapy before documented disease progress10n and subse-
`quently received chemotherapy. Their disease progressed
`subsequent to chemotherapy. We censored these patients at
`the stait of chemotherapy.
`.
`.
`.
`.
`,
`.
`Higher doses of V A did not prolong the time until disease
`progression. The median time to disease progression was 8
`months for patients on the l60-mg arm, 7 months for
`patients on the 800-mg arm, and 8 months for patients on the
`1,600-mg arm. These differences were not of statistical
`significance.
`
`Downloaded from ascopubscrg by 151.194.33.114 on March 28, 2017 from 151.194.033.114
`Copyright © 2017 American Society of Clinical Oncology, All rights reserved.
`
`AstraZeneca Exhibit 2150 p. 4
`
`

`

`68
`
`ABRAMS EI'AL
`
`
`Eble 2. Maximum Response to Megastrol Acetate
`Treatment Arm
`
`
`160 mg
`800 mg
`1,600 mg
`Total
`
`
`
` Response No. % No. % No. % No. %
`
`
`
`
`
`
`
`
`
`
`
`ion—Free
`ProportionProgress
`
`
`o
`
`12
`
`24
`
`36
`
`4a
`
`60
`
`72
`
`84
`
`95
`
`108
`
`Months
`
`from Study Entry
`
`Fig 2. Time to disease progression by treatment arm (160-mg arm: - - -,
`n = 123, median = 8.3; 800-mg arm: ——, n = 124, median = 7.0;
`1,600-mg arm: 7, n = 119, median = 8.1; P: .57).
`
`sion were prior hormone therapy (P : .01), total number of
`metastatic sites (P : .04), estrogen receptor status (P : .04),
`disease-free interval (P : .04), and patient race (P : .03).
`Variables that correlated with better prognoses included no prior
`hormone treatment, fewer metastatic sites, estrogen receptor-
`positive tumors, longer disease-free interval, and white race.
`The following did not correlate univariately with time to
`disease progression:
`treatment arm; whether or not
`the
`patient received prior chemotherapy, RT, or tamoxifen;
`presence or absence of visceral metastases; progesterone
`receptor status; and performance score.
`
`'lable 3. Results of Multivariate Oox Regre$ion Analysis
`Overall Sirvival
`TImeto Regression
`
`(N : 361)
`(N : 361)
`Better
`Better
`
`Variable
`PR
`P
`Prognosis
`RR
`P
`R‘ognosis
`
`1.56
`1.02
`1.31
`
`.0002
`.0001
`.0001
`
`Prior HT
`Age, years
`No. of mets
`Performance
`Lower more
`.10
`1.16
`Lower score
`.0001
`1.46
`more
`
`DFI, years Longer DFI 1.17 .05 Longer DFI 1.14 .11
`
`
`
`
`
`
`No prior HT
`Older
`Fewer mets
`
`1.46
`1.03
`1.10
`
`.0015
`.0001
`.13
`
`No prior HT
`Older
`Fewer mets
`
`NOTE. Prior Ht : no versrsyes age at sudy entry : continuous sale; No.
`of mets : total number of metastatic sites at study entry; performance wore :
`continuous scale; DFI (diagnosis to firs recurrence) : at presentation versus
`within 2 yearsverais 2 2 years after diagnosis
`Abbreviations RR, rid< ratio; HT, hormone therapy; DFI, disease-free interval
`(diagnosisto first recurrence).
`
`123
`2
`121
`13
`15
`67
`23
`2
`1
`
`100
`2
`100
`11
`12
`55
`19
`2
`1
`
`123
`3
`120
`7
`25
`61
`25
`2
`0
`
`100
`2
`100
`6
`21
`51
`20
`2
`0
`
`119
`3
`116
`13
`18
`59
`20
`1
`5
`
`100
`3
`100
`11
`16
`51
`17
`1
`4
`
`365
`8
`357
`33
`58
`187
`68
`5
`6
`
`100
`2
`100
`9
`16
`52
`20
`1
`2
`
`'lbtal patients
`Not assessable
`'lbtal ans-hie
`Complete response
`Partial response
`Sable disease
`Disease progroolon
`Early death
`'lbxicity failure
`Response (complete +
`partial)
`95%oonfidence
`
`16-32% 19-36% 19-36%interval 21-30%
`
`28
`
`23
`
`32
`
`27
`
`31
`
`27
`
`91
`
`25
`
`
`
`
`
`
`
`*XZ : 0.532, 2 df, not significant; Mantel-Haeniel X2 : 0.405, 1 df, not
`significant.
`
`Univariate analysis of multiple pretreatment characteris-
`tics showed that age at study entry (P : .0001) and presence or
`absence of bone metastases (P : .007) correlated most highly
`with time until disease progression. Younger patients were at a
`greater risk of progressing compared \Nlfl’l older patients (risk
`ratio : 1.03). Patients who had bone metastases had a risk of
`progressing that was 37% greater than the risk for those patients
`who did not have bone involvement (risk ratio : 1.37).
`Other variables in this univariate analysis that also
`correlated, although weakly, with time until disease progres-
`
`
`
`
` 0
`
`12
`
`24
`
`36
`
`48
`
`60
`
`72
`
`84
`
`96
`
`108
`
`
`
`ProportionResponding
`
`Months
`
`from Response
`
`Fig 1. Response duration by treatment arm (160-mg arm: - - -, n = 28,
`median = 17.1 ; 800-mg arm: — —, n = 32, median = 13.5; 1,600-mg arm:
`7, n = 31, median = 7.8; P: .0028).
`
`Downloaded from ascopubsorg by 151.194.33.114 on March 28, 2017 from 151.194.033.114
`Copyright © 2017 American Society of Clinical Oncology, All rights reserved.
`
`AstraZeneca Exhibit 2150 p. 5
`
`

`

`MEGESTROL ACEIATE IN MEIAS'IATIC BREAST CANOE?
`
`69
`
`'lable 4. Comparability of CALGB and POA Patient Samples
`CALGB
`FAO
`No.
`
`%
`
`No.
`
`%
`
` P
`
`Value
`
`Demographics
`White
`Age 2 60 years
`Prior treatment received (as many
`asapply)
`Chemotherapy
`RT
`Hormonetherapy
`Sites of metastases measurable
`and/ or amble (as many
`asapply)
`Bone
`Soft tissue
`Viwera
`N umber of metastatic s'tes mea-
`surable and/ or amble
`3 or 4 sites
`2 5 sites
`Pretreatment clinical satus
`Performance more 2 1
`Fbsitive esrogen reoeptor Satus
`Fbsitive progesterone receptor
`status
`Disease-free interval
`
`327
`235
`
`141
`176
`224
`
`229
`366
`150
`
`107
`8
`
`162
`273
`
`209
`
`89
`64
`
`39
`48
`61
`
`63
`100
`41
`
`29
`2
`
`45
`94
`
`79
`
`149
`111
`
`93
`87
`172
`
`126
`172
`65
`
`85
`8
`
`121
`129
`
`82
`
`87
`65
`
`54
`51
`100
`
`73
`100
`38
`
`49
`5
`
`70
`92
`
`68
`
`NS
`NS
`
`.001
`NS
`< .001
`
`.019
`NS
`NS
`
`< .001
`
`.001
`NS
`
`.018
`N S
`
`None, diagnosed with metas-
`16
`27
`1O
`37
`tases
`
`2 2 years 56 229 63 96
`
`
`
`
`Abbreviation: NS, not s'gnificant.
`
`Prior hormone therapy (P : .01), estrogen (P : .03) and
`progesterone (P : .05) tumor status, prior RT (P : .05), and
`disease-free interval (P : .03) also correlated with overall
`survival. Better prognoses were correlated with the follow-
`ing: no prior hormone treatment, no prior RT, estrogen
`receptor-p0sitive/progesterone receptor-positive tumors, and
`longer disease-free interval.
`Treatment arm and whether or not the patient received prior
`chemotherapy and tamoxifen did not correlate with survival.
`Table 3 lists data from the multivariate model for overall
`
`survival. Prior honnone treatment (P : .0002), patient age
`(P : .0001), number of metastatic sites (P : .0001), and
`performance score (P : .0001) correlated strongly with
`overall survival. Disease-free interval was of lesser impor-
`tance (P : .04). Favorable characteristics included the
`following: no prior hormone treatment, older age, fewer
`involved sites> better performance score, and longer disease-
`free interval.
`
`We also examined the relation of several variables
`
`simultaneously with time to progression. Table 3 lists data
`from the resulting multivariate model. Patient age (P :
`.0001) and prior treatment with hormonal agents (P :
`.0015) were strongly associated with time to disease progres-
`sion. Of lesser importance were disease-free interval (P :
`.11), number of metastatic sites (P : .13), and performance
`score (P : .10).
`Favorable characteristics included the following: no prior
`hormone treatment, older age, longer disease-free interval,
`fewer involved sites, and lower performance score.
`
`Overall Survival
`
`Treatment arm did not correlate with survival (Fig 3). The
`median survival from study entry was 28 months (low dose),
`24 months (mid dose), and 29 months (high dose).
`As with time to disease progression, age at study entry
`(P : .0007) and presence or absence of bone metastases
`(P : .0005) correlated most highly with survival in univari—
`ate analysis. In addition, the number of metastatic sites (P :
`.0001) and performance score (P : .0001 ) were of statistical
`significance.
`Again, younger patients had a higher risk of dying earlier
`compared with older patients (risk ratio : 1.02). The
`instantaneous risk of dying for patients with bone involve-
`ment was nearly 50% greater than that of patients without
`bone metastases (risk ratio : 1.49).
`
`
`
`1
`
`. 0 -
`
`
`
`
`2’ 0.8
`
`0 . 6
`
`o . 4-
`
`> > 3
`
`0)
`I:
`
`o I
`
`On
`
`.
`
`o n
`
`. 0.2 -
`
`0.0 - |
`0
`
`|
`12
`
`|
`24
`
`l
`36
`
`l
`48
`
`l
`60
`
`l
`72
`
`|
`84
`
`l
`96
`
`|
`108
`
`Months
`
`from Study Entry
`
`Comparison with POA Study
`
`Fig 3. Overall survival by treatment arm (160-mg arm: - - -, n = 123,
`median = 27.8; 800-mg arm: ——, n = 124, median = 24.4; 1,600-mg
`arm: 7, n = 119, median = 28.6; P: .54).
`
`Table 4 lists the demographic profiles of POA patients
`compared with the CALGB patients.
`In general, POA
`patients were sicker at study enrollment than were the
`
`Downloaded from ascopubsorg by 151.194.33.114 on March 28, 2017 from 151.194.033.114
`Copyright © 2017 American Society of Clinical Oncology. All rights reserved.
`
`AstraZeneca Exhibit 2150 p. 6
`
`

`

`70
`
`ABRAMS EI'AL
`
`
`ProportionSurviving O
`OOO
`
`
`
`b-0’a
`
`M
`
`CALGB patients. On study entry, POA patients tended to
`have more extensive metastatic disease, worse performance
`status, fewer progesterone receptor-positive tumors, and
`more prior chemotherapy than did the CALGB patients.
`About one half of the patients in the POA group had three
`or more metastatic sites, whereas only one third of the
`patients in the CALGB group had three or more sites (P <
`.001). Seventy-three percent of the POA patients had bone
`metastases, compared with 63% of the CALGB patients
`(P : .02). Only 30% of the POA patients had a performance
`score of 0 (normal functioning) compared with 55% of the
`CALGB patients (P : .001). The POA patients had fewer
`progesterone receptor-positive tumors than did the CALGB
`patients (68% V 79%; P : .02). Fifty-four percent of the
`POA patients had prior chemotherapy compared with 39%
`of the CALGB patients (P : .001).
`Within the CALGB study,
`the treatment groups were
`comparable on pretreatment variables; however, within the
`POA study,
`the arms were not always comparable.
`In
`contrast to the POA 800-mg arm, the POA 160-mg arm had a
`higher incidence of visceral metastases (45% V 31%) and
`more sites of metastatic disease (29% V 13% with four sites).
`Thus, the POA patients randomized to the low-dose arm
`were, by chalice, sicker at study entry than were those
`patients assigned to the 800-mg arm (data not shown).
`Table 5 lists tumor responses of patients in the two studies
`by treatment aim. Patients in the low-dose arm of the POA
`study had considerably lower tumor response than did
`patients in any of the other arms. Overall survival did not
`differ by treatment arm in the CALGB or POA study.
`However, CALGB patients experienced longer overall sur-
`vival compared with POA patients (Fig 4). In particular,
`patients in the low-dose POA arm had significantly worse
`survival compared with patients in the low-dose CALGB
`arm (P : .001). The moderate—dose POA and CALGB arms
`did not differ, a finding that indicates that the patients on
`these arms likely had reasonably similar pretreatment char—
`acteristics. Similarly, time to disease progression did not
`differ by study or by treatment arm (data not shown).
`
`Toxicity in CALGB Study
`
`Table 6 lists by treatment arm the maximum toxicities
`reported during MA. The table do