Reprintedfrom
`Volume 15, Number 6
`
`June 1, 1997
`
`JOURNAL OF
`CLINICAL
`ONCOLOGY
`
`Rundomized trial ofeyelophosphamide, methotrexate, und fluorouracil chemotherapy addedto tamoxifen ay udjuvanttherupy In postmenopausal womenwith node-positive estrogen and/or progesterone ceceptor-povitive breast cancer: a report ofthe
`National Cancer Institute ofCanada Clinical Trials Group, Breast Cancer Site Group,
`
`ASCO_JCO
`
`Disclaimer
`The ideas and opinions expressed in the Journal ofClinical Oncology (JCO) do not necessarily reflect those of the American Society
`of Clinical Oncology (ASCO). The mention of any product, service, or therapy in any JCOarticle should not be construed as an
`endorsementof the products mentioned.
`
`It is the responsibility of the treating physician or other health care provider, relying on independent experience and knowledgeof the
`patient, to determine drug dosages and the best treatmentfor the patient. Readers are advised to check the appropriate medical
`literature and the productinformation currently provided by the manufacturer of each drug to be administered to verify the dosage,
`method, and duration of administration, or contraindications. Readers are also encouraged to contact the manufacturer with questions
`aboutthe features or limitations of any products. ASCO assumesno responsibility for any injury or damageto personsor property
`arising out of or related to any use of the material contained in JCO orto any errors or omissions.
`
`ASC@)
`
`American Society of Clinical Oncology
`
`Making a worldofdifference in cancer care
`
`Genentech 2132
`Genentech 2132
`Celltrion v. Genentech
`Celltrion v. Genentech
`IPR2017-01122
`IPR2017-01122
`
`

`

`Randomized Trial of Cyclophosphamide, Methotrexate,
`and Fluorouracil Chemotherapy Added to Tamoxifen as
`Adjuvant Therapy in Postmenopausal Women With
`Node-Positive Estrogen and/or Progesterone Receptor-
`Positive Breast Cancer: A Report of the National Cancer
`Institute of Canada Clinical Trials Group
`
`By Kathleen |. Pritchard, Alexander H.G, Paterson, Sheldon Fine, Nancy A. Paul, Benny Zee, Lois E. Shepherd,
`Hakam Abu-Zahra, Joseph Ragaz, Margaret Knowling, Mark N. Levine, Shail Verma, Daniele Perrault,
`P.L. David Walde, Vivien H.C. Bramwell, Mate Poljicak, Norman Boyd, David Warr, Brian D. Norris, David Bowman,
`George R. Armitage, Harold Weizel, Robert A. Buckman, and the National CancerInstitute of Canada Clinical Trials
`Group Breast Cancer Site Group
`
`Purpose and Methods: By the mid 1980s, tamoxifen
`alone was considered standard adjuvant therapy for
`postmenopausal women with node-positive, estrogen
`receptor (ER)- or progesterone receptor (PgR)-positive
`breast cancer. From 1984 through 1990, 705 eligible
`postmenopausal women with node-positive, ER- or PgR-
`positive breast cancer were randomized te a National
`CancerInstitute of Canada Clinical Trials Group (NCIC
`CTG) study that compared tamoxifen 30 mg by mouth
`daily for 2 years (TAM) versus TAM plus chemotherap
`with all-intravenous cyclophosphamide 600 mg/m’,
`methotrexate 40 mg/m’, and fluorouracil 600 mg/m?
`given every 21 daysfor eight cycles (CMF}.
`Results: There were no significant differences in over-
`all survival, recurrence-free survival, locoregional recur-
`
`rence-free survival, or distant recurrence-free survival
`between the two treatment arms. However, there was
`significantly greater severe toxicity, which included ley-
`kopenia (P < .0001), nausea and vomiting [P < .0001),
`and thromboembolic events (P < .0001), as well as sig-
`nificantly more mild or greater toxicity, which included
`thrombocytopenia (P = .04), anemia (P = .02), infection
`(P = .0004), mucositis (P = .0001), diarrhea (P = .0001),
`and neurologic toxicity (P = .006), in women who re-
`ceived TAM plus CMF.
`Conclusion: The addition of CMF to TAM adds no ben-
`efit and considerable toxicity in this group of women.
`J Clin Oncol 15:2302-2311.© 1997 by American So-
`ciety of Clinical Oncology.
`
`Y THE MID 1980s, chemotherapy had becomestan-
`dard adjuvant therapy for node-positive premeno-
`pausal women,‘? while tamoxifen was the usual recom-
`mendation for many postmenopausal women with breast
`cancer.*** Tamoxifen’s utility in women with tumors to-
`tally negative for csuogen receptor (ER) and progesterone
`receptor (PgR) remained controversial,*** andits role in
`node-negative women was not established at that time.’
`However, it was, and still is by many today, considered
`
`the treatment of choice for node-positive postmenopausal
`women with ER- or PgR-positive breast cancer following
`primary therapy with surgery with or without breast irra-
`diation>’ The role of cytotoxic therapy in postmeno-
`pausal women was controversial in the mid 1980s and
`has remained so,’* despite more recent suggestionsofits
`efficacy.> ?
`in
`Thus,
`in 1984, we began the trial reported here,
`which node-positive, ER- or PgR-positive women follow-
`
`
`
`From the Toronto-Sunnybrook Regional Cancer Centre, Ontario
`Cancer Treatment and Research Foundation (OCTRF), University
`of Toronto; Ontario Cancer Institute; Toronto Hespital, Toronto;
`Tom Baker Cancer Centre; University of Calgary, Calgary; Credit
`Valley Huspital, Mississauga; National Cancer Institute of Canada
`Clinical Trials Group Central Office, Queen’s University, Kingston;
`Windsor Regional Cancer Centre, OCTRF, Windsor; British Colum-
`bia Cancer Agency and University of British Columbia, Vancouver;
`Hamilton Regional Cancer Centre, OCTRF; McMaster University,
`Hamilton; Ottawa Regional Cancer Centre, OCTRF; University of
`Ottawa and Ottawa Civic Hospital, Ottawa; Plummer Memorial
`Hospital, Sault Ste Marie; London Regional Cancer Centre, OCTRF;
`University of Western Ontario, London; Hépital Notre-Dame, Uni-
`versity of Montréal, Montreal; British Columbia Cancer Agency;
`Fraser Valley Cancer Centre, Surrey; Health Sciences Centre, Uni-
`
`versity of Manitoba; Manitoba Cancer Foundation, Winnipeg; Sas-
`katoon Cancer Centre, Saskatchewan Cancer Foundation; Univer-
`sity of Saskatchewan, Saskatoon, Canada.
`Submitted July 11, 1996; accepted February 13, 1997.
`Supported by the National Cancer Institute of Canada, To-
`ronto,
`the Medical Research Council of Canada, Ottawa, and
`Zeneca Pharma Inc (formerly ICI Pharma), Mississauga, Can-
`ada,
`Address repprint requests to Kathleen I. Pritchard, MD, Head,
`Medical Oncology/Haematology, Toronto-Sunnybrook Regional
`Cancer Centre and Sunnybrook Health Science Centre, University
`of Toronto, 2075 Bayview Ave, Toronto, Ontario, Canada M4N 3MS5;
`Email kathy_pritchard@ octrf.on.ca.
`© 1997 by American Society of Clinical Oncology.
`0732-183X/97/1 506-0006$3.00/0
`
`2302
`
`Journal of Clinical Oncology, Vol 15, No 6 (June}, 1997: pp 2302-2311
`
`
`
`

`

`CMF ADDED TO POSTMENOPAUSAL ADJUVANT TAMOXIFEN
`
`2303
`
`ing primary surgery with or without breast irradiation
`were randomly allocated to receive tamoxifen 30 mg daily
`for 2 years (TAM) or TAM plussix cycles of chemother-
`apy with intravenous cyclophosphamide, methotrexate,
`and fluorouracil (CMF), a standard cytotoxic regimen of
`the time. We have previously published preliminary re-
`sults from thistrial, including a report of an unexpectedly
`bigh level of thromboembolic events in the TAM plus
`CMFarm'*!° and preliminary results of the major end
`points, disease-free and overall
`survival,
`in abstract
`form.’*"’ However, this represents the first full report of
`the major outcomesof the trial, which completed accrual
`in December 1990.
`
`surgery. Approval ofthe trial protocol by a local human investiga-
`tions comumittee was performed in each center and in the data coordi-
`nating center.
`
`Treatment Regimens
`Tamoxifen was given in a dosage of 30 mg by mouth daily for
`2 years, while CMF was given concurrently with tamoxifen, all
`intravenously in doses of cyclophosphamide 600 mg/m?, methotrex-
`ate 40 mg/m’, and fluorouracil 600 mg/m? body-surface area every
`21 days for 8 cycles. CMF doses were reduced to 50%if the neutro-
`phil count was less than 1,500/pL but = 1,000/uL, or the platelet
`count was less than 100,000/uL but 2 75,000/uL at the time that
`course was due. If the neutrophil count was less than 1,000 or the
`platelet count less than 75,000, no chemotherapy was given until
`both counts reached acceptable levels.
`
`PATIENTS AND METHODS
`
`Statistical Methods
`
`Study Design
`This study is a randomized multicenter clinical trial designed and
`performed by the National CancerInstitute of Canada Clinica] Trials
`Group (NCIC CTG) Breast Cancer Site Group (BCSG). The fourth
`in a series of trials in women with breast (mammary) cancer,it is
`therefore called MA.4. Randomization was performed by telephon-
`ing the NCIC CTG data coordinating center. Patients were stratified
`according to (1) method of primary treatment(modified radical mas-
`tectomy [MRM], partial mastectomy [PM], or PM plusbreast radia-
`tion); (2) numberof involved axillary nodes (oneto three v = four);
`(3) time since menopause (< 5 years v = 5 years); and (4) level of
`receptor assay (both between 10 and 30 fmol/mgcytosol, either ER
`or PgR = 30 fmol/mg). Patients were randomly allocated with equal
`probability to one of the two treatment arms with TAM or TAM
`plus CMFusing a blocked randomization procedure. A samplesize
`of SOO patients wasoriginally calculated to give an 80% powerto
`detect a hazardsratio of 1.5 with respect to overall survival using
`a one-sided 5%-level test, but this sample size was subsequently
`increased to 700 patients to detect a hazards ratio of 1.38 under the
`same conditions.
`
`Patient Eligibility
`Women were required to be postmenopausal (last menstrual period
`at least 6 months before surgery, both ovaries removed, or hysterec-
`tomy and age > 50 years) and to have had the following: (1) com-
`plete removal of the primary tumorwith histologically clear margins
`and at least a level 2 axillary node clearance (up to the axillary
`vein); (2) histologic examination of at least four axillary nodes; (3)
`at least one axillary node microscopically involved with breast can-
`cer; and (4) ER or PgR = 10 fmol/mgof tissue cytosol. Metastases
`were excluded by a panel ofliver function tests, serum calcium and
`alkaline phosphatase levels, carcinoembryonic antigen (CEA)level,
`chest radiograph, and nuclear bone scan with radiographs of suspi-
`cious areas, Patients with clinical stages T1-3NO-1MO were consid-
`ered eligible. Criteria of ineligibility included a serum creatinine
`level greater than 1.5 mg/100 mL, and any previous malignancy
`excepting basal cell or squamous cell carcinoma of the skin, or
`carcinoma of the cervix, thyroid, uterus, or colon, treated and pre-
`sumed cured more than 5 years previously. No radiation was permit-
`ted except to the partially removed breast. Women were required to
`be in generally good health apart from the diagnosis of breast cancer
`and to be randomized and to begin therapy within 10 weeks of
`
`All eligible patients were included in this analysis. The data set
`for the final analysis was frozen on April 15, 1994. Recurrence-free
`survival, locoregional recurrence-free survival, distant recurrence~
`free survival, and overall survival were chosen as the major study
`end points. Recurrence-free survival was assessed using time of
`first recurrence as an end point. In assessing patterns of recurrence,
`recurrences within 30 days were considered simultaneous. An analy-
`sis of first breast recurrence was performed on the 247 patients who
`had received partial mastectomy with or without breast radiation.
`Sites of distant recurrence were further subdivided into first bone
`recurrence, first lung recurrence, and first liver recurrence. When
`two or more recurrences were determined simultaneously, the more
`prognostically serious site was taken to be thesite offirst recurrence.
`The order used was breast, chest wall/nodes, and distant (bone,lung,
`and liver), Dating of recurrence was based on thefirst date of onset
`of a sign, but never of a symptom. The date of first detection of a
`palpable lesion was acceptable only when the diagnosis of tumor
`involvement was subsequently established. The diagnosis of recur-
`rent disease by radiographs or scans was also dated from the first
`positive record, even if that was determined in retrospect.'*
`Overall survival and recurrence-free survival rates were analyzed
`by the Kaplan-Meier method.’” Log-rankstatistics were used to test
`the difference between the two treatment arms. Multivariate analyses
`of prognostic factors for each of the major end points were performed
`and incorporateda variety of patient and tumorcharacteristics, which
`included the following: (1) center accrual size (<= 25 v > 25 pa-
`tients); (2) age (continuous variable [cont]); (3) Kamofsky perfor-
`mance status (cont); (4) number of years since menopause (cont):
`(5) prior hormonereplacement therapy (never v received > 6 months
`before study entry v received unti] < 6 months before study entry);
`(6) method of primary weatment (MRM v PM v PM plus breast
`radiation); (7) clinical stage (1 v I v ID); (8) pathologic stage (I v
`TID; (9) receptor levels (both < 30 fmol/mg v one = 30 fmol/mg);
`(10) nodal status (oneto three v = four positive); (11) height (cont);
`(12) weight (cont); and (13) auxometric index, a clinical index of
`rate of growth of the tumor, which was based on an assessment of
`clinical progression over time (no breast change and duration > 12
`weeks v no breast change and duration -«< 12 weeks v breast change
`and duration > 12 weeks v breast change and duration < 12
`weeks).*° 4 Cox regression mode] was used to assess prognostic
`variables.?' Treatmenteffect after adjustmentfor important prognos-
`tic factors was estimated within the final Cox model. The propor-
`tional hazards assumption was validated by plots of cumulative haz-
`
`
`
`
`
`
`
`

`

`2304
`
`PRITCHARD ET AL
`
`Table 1. Patient Characteristics
`Treatment Arm
`TAM (n = 352)
`TAM + CMF (n = 353}
`Both {n = 705}
`No
`%
`No
`%
`Factor
`No
`%
`
`
`224
`32
`96
`
`166
`186
`
`73
`101
`140
`32
`6
`
`207
`118
`27
`
`12
`142
`198
`
`Nn
`92
`238
`11
`
`14
`299
`39
`
`63 6
`91
`273
`
`472
`§28
`
`207
`287
`39 8
`91
`17
`
`58 8
`335
`OF
`
`34
`40 3
`563
`
`31
`262
`67.6
`3)
`
`40
`849
`WW]
`
`234
`32
`87
`
`174
`179
`
`76
`91
`15]
`31
`4
`
`214
`106
`33
`
`15
`162
`176
`
`7
`99
`224
`23
`
`8
`316
`29
`
`663
`9]
`246
`
`493
`507
`
`21.5
`258
`A28
`&8
`VW
`
`606
`300
`94
`
`42
`45 9
`499
`
`20
`280
`635
`65
`
`23
`89 5
`B2
`
`Primary surgery*
`MRM
`PM
`PM + radiation
`Receptor slatus”
`ER ond PgR < 30 fmol/mg
`ER or PgR = 30 fmol/mg
`Years postmenopausal*
`s§
`5-10
`10-20
`20-30
`= 31
`No of positive nades”
`1-3
`49
`210
`No of nodes examined
`4
`5-9
`210
`Clinical stage
`Unknown
`|
`it
`Ml
`Pathologic stage
`Unknown
`ll
`Mt
`Auxometric classiheation
`101
`71
`77
`27
`125
`44
`No change > 12 weeks
`508
`358
`521
`184
`49 4
`174
`No change = 12 weeks
`201
`142
`198
`70
`205
`72
`Change > 12 weeks
`
`
`
`20 4 13472 190
`Change < 12 weeks
`62
`176
`“Stratified before randomization
`
`458
`64
`183
`
`340
`365
`
`149
`192
`291
`$3
`10
`
`A2l
`224
`60
`
`27
`304
`374
`
`18
`191
`462
`34
`
`22
`615
`68
`
`650
`91
`259
`
`48 3
`517
`
`21.1
`272
`Al3
`8°
`14
`
`597
`318
`85
`
`28
`43]
`531
`
`26
`27 1
`655
`48
`
`31
`87 2
`91
`
`ards functions Treatment covanate interacuons were also tested
`under the Cox model
`Toxicity data were collected using the Eastern Cooperative Group
`(ECOG) toxicity cntena and analyzed using x? tests
`
`RESULTS
`
`Randomization and Eligibility
`Seven hundred thirty-six women were randomized to
`one of the two treatment arms between January 1984 and
`December 1990. Of these 736 women, 31 (4.2%) were
`meligible for reasons that included incomplete surgery (n
`= 11), metastatic disease (1 = 4), madequate staging
`information (n = 3), involved margins (n = 3), receptor
`status unknown or negative (n = 3), clinical stage T4 (n
`= 2), premenopausal status (n = 2), abnormal renal func-
`
`tion (n = 1), randomized more than 10 weeks after sur-
`gery (n = 1), and other serious illness (n = 1). Of the
`remaining 705 ehgible women, 352 were randomized to
`TAM alone and 353 to TAM plus CMF.
`Patient Characteristics
`
`The two treatment arms were comparable according
`to their baseline characteristics (Table 1). Patients were
`stratified according to numbers of years postmenopausal,
`primary surgery, receptor status, and numberofpositive
`nodes before randomization, and so the distribution of
`these factors was well balanced between the two treat-
`ment arms. Other potential prognostic factors, for which
`there was nostratification before randomization, are also
`well balanced between the two treatment arms (Table 1).
`
`
`
`
`
`
`
`

`

`CMF ADDED TO POSTMENOPAUSAL ADJUVANT TAMOXIFEN
`
`2305
`
`Table 2. Toxicities by Treatment Arm
`Treatment Arm
`
`TAM + CMF
`TAM (n = 352}
`(a = 353)
`
`Toxicity/Grade
`No.
`%
`No.
`%
`Leukopenia*
`1
`2
`3
`4
`Thrombocytopeniat
`1
`2
`3
`Anemia?
`1
`2
`3
`HemorrhageS
`1
`2
`Infection]
`1
`2
`3
`4
`Vascular!
`1
`2
`3
`4
`5
`Nausea & vomiting**
`]
`2
`3
`Mocositistt
`1
`2
`3
`
`19
`0
`0
`0
`
`1
`0
`1
`
`0
`9
`Go
`
`0
`9
`
`i
`4
`1
`0
`
`0
`4
`q
`4
`0
`
`24
`6
`0
`
`0
`0
`0
`
`5.4
`0
`0
`0
`
`0.3
`0
`0.3
`
`Go
`o
`0
`
`0
`0
`
`0.3
`1.1
`0.3
`0
`
`0
`11
`0.3
`Tal
`0
`
`6.8
`a
`0
`
`0
`0
`0
`
`43
`80
`91
`7?
`
`24
`7
`1
`
`40
`é
`T
`
`4
`]
`
`20
`17
`3
`4
`
`2
`12
`24
`7
`3
`
`84
`135
`39
`
`39
`22
`4
`
`122
`227
`25.8
`22.4
`
`6.8
`2.0
`0.3
`
`11.3
`1.7
`0.3
`
`1.1
`0.3
`
`5.7
`4.8
`09
`1.13
`
`0.6
`3.4
`6.8
`2.0
`0.8
`
`23,8
`38.2
`11.7
`
`TT
`6.2
`1.1
`
`Treatment Compliance
`The majority (84%) of women discontinued tamoxifen
`either at the end of 2 years (69.5%) or at disease recur-
`rence (14.7%), as directed by the protocol. Only 4% dis-
`continued tamoxifen because of toxicity: 5.6% in the
`TAM plus CMF arm and 2.3% in the TAM-alone arm.
`Forty-nine (7.0%) werestill receiving tamoxifen,in viola-
`tion of the protocol, at the time of this analysis. Of these,
`19 were in the TAM plus CMFarm (5.6%) and 30 were
`in the TAM-alone arm (8.7%) (P = .138).
`Two hundred ninety-six women (84%) completed eight
`cycles of therapy. Ninety-three percent of patients had
`received at least 85% of the correct dose, taking into
`consideration the dose modifications described in the pro-
`tocol (predicted dose). However, compliance with chemo-
`therapy was inferior in the early part of the study. The
`proportions of patients who received less than 85% of
`the total predicted dose for eight cycles in 1984, 1985,
`and 1986 were 19%, 11%, and 10%, respectively, com-
`pared with less than 5% from 1987 to 1990. However,
`overall survival and recurrence-free survival rates for pa-
`tients entered between 1984 and 1986 versus those en-
`tered between 1987 and 1990 were notsignificantly dif-
`ferent (P = .54; P = .58&). A Cox regression model on
`the rank of the date of entry for patients was used to
`assess the effect of early noncompliance problems. They
`were not significant with respect to overall and/or recur-
`rence-free survival (P = .51; P = .60).
`
`Toxicity
`
`Table 2 lists worst toxicities by treatment arms. Patients
`in the TAM plus CMF arm hada significantly higher
`incidence of severe or worse (grade =: 3) toxicities, which
`included leukopenia (48% v 0%; P < .0001), nausea and
`vomiting (11% v 0%; P < .0001), and vascular events
`(10% v 1%; P < .0001). Significantly higher incidences
`of more than mild (grade = 2) thrombocytopenia (P =
`.04), anemia (P = .02), infection (P = .0004), mucositis
`(P = .0001), diarrhea (P = .0001), and neurologic toxicity
`(P = .006) were also found in the TAM plus CMF arm.
`There wasalso significantly more hemorrhage (P = .06)
`and renal damage (P = .03) of any grade in the TAM
`plus CMF amm.Deaths due to toxicity alone included one
`from pulmonary embolism and one from a cerebrovascu-
`lar accident (CVA). Both occurred in patients randomized
`to TAM plus CMF,although one patient never actually
`received CMF, while one received only one course of
`CMFand did not develop the CVA until 14 months later.
`Both werestill receiving tamoxifen at or within days of
`the time of the adverse event. Deaths related to toxicity
`and to breast cancer involved the occurrence of a deep-
`
`“P< .0001 (0-1 v 2+ or 0-2 v 34).
`TP = 04 (0-1 v 2+).
`$P = .02 (0-1 v 2+); P= .0001 (0 v 1-3).
`§P = .06 (0 v 1+).
`QP = 0004 (0-1 v 2+}; P = .069 (0-2 v 3+}.
`"P< 0001 (0-1 v 2+ or 0-2 ¥ 34).
`“*P < 000] (0-] y 2+ or 0-2 v 3+].
`ttP < .0001 (0-1 v 2+).
`
`vein thrombosis followed by a gastrointestinal hemor-
`rhage in one patient and a CVAin another. Both patients
`were randomized to the TAM plus CMF arm,but both
`had completed CMF = 1 year before the adverse event
`occurred. The patient who developed a CVA wasstill
`receiving tamoxifen at the time of the adverse event,
`while the other had discontinued tamoxifen a few months
`
`earlier. Twenty-eight second malignancies were reported
`
`
`
`
`
`

`

`2306
`
`PRITCHARD ET AL
`
`Table 3. Secand Malignancies
`Treatment Arman
`TAM
`TAM -+ CMF
`Total
`
`%
`No.
`x
`Site
`No.
`%
`Na.
`
`92
`None
`325
`92
`326
`651
`92
`3
`Breast
`12°
`4
`10
`22°
`3
`09
`Skin
`5
`14
`3
`8
`1.
`0.5
`Lung
`2
`0.5
`2
`4
`0.5
`0.2
`Colorectal
`3
`0.5
`1
`4
`0.5
`0?
`Ovary
`0
`9
`3
`3
`0.4
`0.5
`Bladder
`0
`0
`2
`2
`0.3
`0.5
`Endometrial
`1°
`0.3
`2
`3°
`03
`0.5
`Biliory Tree
`0
`0
`2
`2
`0.3
`0.5
`Other
`5
`1.4
`2
`7
`0.9
`Total
`
`353352 708
`
`*Onepatientfirst developed a second breast carcinoma and then an endometrial carcinoma,
`
`who did not receive PM plus radiation (ic, those who
`received either mastectomy or PM without radiation) had
`an inferior recurrence-free survival (P = .0085). Auxome-
`tric index was significant in univariate analysis, but not
`in the Cox regression model. Treatment effect of TAM
`plus CMF versus TAM alone was not significant after
`adjusting for these prognostic factors (P = .59). Thetests
`for interaction between treatment and other prognostic
`factors were not significant.
`
`Breast and Locoregional Recurrence-Free Survival
`
`in the TAM and 27 in the TAM plus CMFarms. Ofthese,
`12 and 10 were second breast primary tumors, while 16
`and 17 were other new primary tumors in the TAM and
`TAM plus CMF arms, respectively. Further details of
`these second primary tumorsare listed in Table 3.
`
`Recurrence-Free Survival
`
`There were 143 (41%) and 142 (40%) recurrences in
`the TAM plus CMF and TAM-alone arms, respectively.
`The patterns ofsites of first recurrence, classified as pre-
`viously described (Methods),are listed in Table 4. They
`are not different between weatment arms. The relative
`risk of recurrence for the TAM plus CMF compared with
`the TAM-alone arm was 0.97 (95% confidence interval
`{CI}, 0.77 to 1.23). The 5-year recurrence-free survival
`rates for the TAM plus CMF and TAM-alone arms were
`64% and 61% (3% difference; 95% CI, -5% to 11%).
`Kaplan-Meier curves for recurrence-free survival are
`shownin Fig 1. There is no significantdifference between
`the two arms (P = 0.80).
`Cox proportional hazards model showed that younger
`patients (P = .0096), patients with four or more involved
`nodes (P = .0001), patients with both receptor levels
`between 10 and 30 fmol/mg (P = .0001), and patients
`
`Thepatterns of recurrence listed in Table 4 demonstrate
`no statistically significant difference in numbers of breast
`recurrences or of locoregional recurrences between treat-
`ment arms. First breast recurrences, of course, can only
`be seen in the patients who received either PM or PM
`plus breast radiation as primary therapy. There were 1]
`and 17 first breast recurrences and 53 and 59first locore-
`gional recurrences in the TAM plus CMF and TAM-
`alone arms, respectively, if we include those determined
`simultaneously with other types of recurrence. After cor-
`recting for simultaneous recurrence in a competing-risk
`model as outlined earlier (see Methods), the TAM plus
`CMF and TAM-alone arms had six and 1] first breast
`recurrences and 43 and 45 first locoregional recurrences,
`respectively. The Kaplan-Meier curves that compared
`Table 4. First Recurrences by Treatment Arm
`first breast recurrence (P = .309) and those that compared
` Treatment
`first locoregional recurrence (P = .72) were not signifi-
`TAM
`TAM + CMF
`cantly different. The relative risk offirst breast recurrence
`
`{n = 352) (n = 353)Site Ppeei
`
`
`for TAM plus CMF versus TAM alone was 0.60 (95%
`ST]
`§
`Breast
`31
`CI, 0.22 to 1.62). The relative risk of first locoregional
`Locoragional
`45
`43
`92
`recurrence was 0.93 (95% CI, 0.61 to 1.41) for TAM plus
`Distont
`86
`94
`73
`CMFversus TAM alone.
`A Cox proportional hazards model showed that patients
`
`
`
`142 143Total -80i
`
`
`
`
`
`

`

`CMF ADDED TO POSTMENOPAUSAL ADJUVANT TAMOXIFEN
`
`
`
`PercentageAlive
`
`Fig 1. Kaplan-Meier curves
`recurrence-free survival by
`of
`treatment arm (P = .80) (TAM,
`++"; TAM plus CMF, -~).
`
`0.0
`g53
`352
`
`1.0
`351
`350
`
`2307
`
`2.0
`340
`336
`
`30 40 50 60 7.0 8.0 9.0 10.0
`315
`242
`178
`112
`7
`36
`6
`0
`312
`236
`#165
`#18
`«77
`41
`7
`o
`
`Time (Crears)
`# At Risk (Arm CT)
`# At Risk (Arm T)
`
`whoreceived breast radiation had a lower risk of first breast
`recurrence than those who were treated with PM alone (P =
`032). The treatment effect of TAM plus CMF versus TAM
`alone on first breast recurrence was not significant after ad-
`justment for other confounding factors (P = .22). A Cox
`proportional hazards model for first locoregional recurrence
`showed that younger age (P = .02), four or more involved
`axillary nodes (P = .0001), both receptor levels between 10
`and 30 fmol/fmg (P = .03), and pathologic stage I tumors
`(P = .003) were each associated with a higher risk of locore-
`gional recurrence. Patients who had PM plus breast radiation
`were also significantly less likely to develop locoregional
`recurrence (P = .02). Treatment effect of TAM plus CMF
`versus TAM alone was not significant for locoregional recur-
`rence after adjusting for significant prognostic factors within
`the final Cox model (P = .83). The tests for interaction
`between treatment and other prognostic factors were not sig-
`nificant.
`
`Distant Recurrence-Free Survival
`
`Distant recurrence-free survival was not significantly
`different between the two treatment arms. The relative
`risk of distant recurrence for TAM plus CMEversus TAM
`alone was 1.05 (95% CI, 0.79 to 1.41). A Cox regression
`model showed that four or more involved axillary nodes
`(P = .0001) and low receptor levels (P = .0001) were
`significantly associated with a higherrisk of distant recur-
`rence. Patients with PM plus breast radiation had a sig-
`
`nificantly lower risk of developing a distant recurrence
`(P = .023). Systemic treatment effect was not significant
`after adjusting for significant prognostic factors (P = .93).
`There were a total of 146 bone recurrences, 60 lung
`recurrences, and 7] liver recurrences, if we include all
`recurrences even if they occurred simultaneously at multi-
`ple sites. Correcting for simultaneous recurrences in the
`competing-risks model, as outlined earlier (see Methods),
`the numbersoffirst bone, first lung, and first liver metas-
`tases become 106,28, and 28, respectively. A Cox propor-
`tional hazards model for first bone recurrence showed
`that four or more involved axillary nodes (P = .0001),
`low (both between 10 and 30 fmol/mg) receptor levels
`(P = .02), and greater patient height (P = .007) were
`significantly associated with a higher risk of first bone
`recurrence. Four or more positive nodes and low receptor
`levels were each significantly associated with a higher
`tisk of developing either first lung or first liver recur-
`rences. Treatmenteffects of TAM plus CMFversus TAM
`alone were not significant for first bone, lung, or liver
`recutrence after adjusting for significant prognostic fac-
`tors (P = .758, P = .84, and P = .34, respectively). The
`tests for interaction between treatment and other prognos-
`tic factors were not significant.
`
`Overall Survival
`
`There were 172 (24%) deaths, 86 in each treatment
`arm,at the timeof this analysis. There was no significant
`
`
`
`

`

`2308
`
`PercentageRecurrence—Free NOo
`hD@Oo°oOo
`
`
`
`of overall survival by treatment
`crm (P = .94} (TAM, ---; TAM
`plus CMF, —}.
`
`PRITCHARD ET AL Fig 2. Kaplan-Meier curves
`
`0.0
`353°
`352
`
`1.0
`330
`319
`
`2.0 3.0 40 5.0 6.0
`306
`264
`200
`132
`280
`288
`254
`192
`120
`94
`
`7.0 8.0 3.0 10.0
`53
`26
`4
`°
`s6
`30
`s
`o
`
`Time (Years)
`a& At Risk (Arm CT)
`At Risk (Arm T
`
`II waspredictive of locoregional recurrence, but not of
`overall survival, distant recurrence, or recurrence in gen-
`eral. Patients who had PM plusbreastradiation were less
`likely than those who had PM or MRMto develop any
`recurrence (P = .009), locoregional recurrence (P = .02),
`or distant recurrence (P = .02), particularly bone recur-
`rence, but this factor was not predictive for lung orliver
`recurrence or for overall survival. Breast recurrence was
`affected only by the use of breast radiation (P = .032).
`
`difference between treatment in the proportion of deaths
`from disease, from complications of disease and treat-
`ment, from second malignancies, or from other causes,
`Kaplan-Meier survival curves (Fig 2) are not significantly
`different between the two arms (P = .94). The relative
`risk of death for women in the TAM plus CMF arm
`versus those who received TAM alone was 1.01 (95%
`CI,0.75 to 1.36). The 5-year survival rates for TAM plus
`CMFversus TAM alone were 82% and 80%,respectively
`(2% difference; 95% CI, -4% to 8%). Cox proportional
`hazards model showed that women with four or more
`nodes involved (P = .0001) and low (between 10 and 30
`An early analysis of data from this trial suggested a
`fmol/mg) receptor levels (P = .0001) hadasignificantly
`borderline improvementin recurrence-free survival, par-
`inferior overall survival] rate. Auxometric index was sig-
`ticularly locoregional recurrence-free survival, with the
`addition of CMF to tamoxifen in this group of women.'®
`nificant in univariate analysis (P = .011), but not in Cox
`However, with more mature follow-up data, even this
`model after controlling for other confounding factors (P
`= .088). The tests for interaction between treatment and
`advantage was no longer seen.’” In this,
`the first full
`other prognostic factors were not significant.
`analysis of the results of this study, we find no advantage
`In summary, the addition of CMF to TAM was not
`in overall] survival, progression-free survival, or locore-
`gional or distant progression-free survival.
`significantly beneficial in terms of recurrence-free sur-
`Many studies now support the utility of tamoxifen as
`vival, overall survival, or any subset of recurrence-free
`adjuvanttherapy in this subset of women,”*!°''?8 as they
`survival (breast, locoregional, or distant recurrence-free
`did when we began this trial in 1984.>° However, the role
`survival). In the Cox model, four or more nodes and low
`of chemotherapyin this setting has remained controversial.
`receptor levels were predictive of recurrence;
`locore-
`Used alone, it appears to have some benefit, as summarized
`gional recurrence; distant recurrence;
`liver,
`lung, and
`bone recurrence; and of overall survival. Young age was
`in the results of the Early Breast Cancer Trialists’ over-
`view!" Tn that analysis, prolonged chemotherapy produced
`predictive of recurrence and of locoregional recurrence,
`a 13% reduction in the annual odds of death (nonsignificant)
`but not of distant recurrence or survival. Pathologic stage
`
`DISCUSSION
`
`
`
`

`

`CMF ADDED TO POSTMENOPAUSAL ADJUVANT TAMOXIFEN
`
`2309
`
`and a 24% reduction in the annual odds of recurrence (sig-
`nificant) in women = 50 years of age. In the same age
`group, tamoxifen produced a reduction of 20% in the annual
`odds of death and of 27% in the annual odds of recurrence
`(both significant). Both of these tamoxifen effects increased
`with increasing ER positivity."
`Trials of chemohormonal therapy in postmenopausal
`womenhave had three possible designs. Some, such as
`our own, compared tamoxifen versus tamoxifen plus che-
`motherapy,”’”’> sometimes with the addition of a no-
`treatment control arm.”'* Some compared chemotherapy
`versus the same chemotherapy plus tamoxifen,?”** while
`others compared tamoxifen alone versus chemotherapy
`alone versus tamoxifen plus the same chemotherapy,"
`with*>** or without®”** a no-treatment control arm.
`Trials of tamoxifen added to chemotherapy versus che-
`motherapy alone in the postmenopausal, node-positive,
`ER-positive group, generally demonstrate, as does the
`overview,'°"! a statistically significant improvement in
`disease-free and overall survival with the addition of ta-
`moxifen.?** However, trials of chemotherapy added to
`tamoxifen versus tamoxifen alone show muchless effect.
`Since the overview analysis of these trials describes a
`significant 27% reduction in the annual odds of recur-
`rence, but no significant overall survival advantage," it
`is not surprising that the individual trials, most of which
`are summarized in the overview data, show a similar but
`more variable picture. Several, like our own, show no
`recurrence-free or overall survival advantage,**”"* others
`demonstrate a significant improvementin recurrence-free
`butnotoverall survival,?’** while some showa significant
`advantage for both.”
`The first of the trials that demonstrated an advantage in
`both disease-free and overall survival was, unfortunately,
`confounded by the fact that both of its treatment arms
`contained prednisone.*"? Interestingly, with regard to
`overall survival, there was little difference between the
`TAM plus prednisone (PT) and no-treatment control
`arms, while the arm in