`
`
`
`Annals of Oncology 3: 611-617, 1992.
`© 1992 Kluwer Academic Publishers. Printed in the Netherlands.
`
`Responseafter withdrawal of tamoxifen and progestogens in advanced breast
`cancer
`
`A. Howell, D. J. Dodwell, H. Anderson & J. Redford
`From the CRC Department of Medical Oncology, Christie Hospital, Manchester and the Department of Surgery, University Hospital of
`South Manchester
`
`Summary. Tumor response after withdrawal of endocrine
`therapy for advanced breast cancer with estrogens and
`androgens is well described. There have been few reports of
`withdrawal responses (WRs)after cessation of treatment with
`the newer antiestrogens and progestogens. We assessed WR
`in women after cessation of adjuvant therapy at first relapse,
`and after progression onfirst, second or third line endocrine
`therapy for advanced disease. One of seven patients (14%)
`responded after cessation of tamoxifen adjuvant therapy at
`relapse. Sixty-five of 72 patients were evaluable for WR after
`cessation of tamoxifen as first line therapy for advanced dis-
`ease. There were five partial responses (8%) and 14 (22%)
`‘no change’ with a median duration of WR of 10 months
`
`(range 3-40 months). WR were seen mainly in soft tissue dis-
`ease but there were two responses in lung and twoin bone.
`Four of 21 (19%) patients had a WR after cessation of
`norethisterone acetate (3) and tamoxifen (1), all used as
`second line therapy. WR are therefore demonstrable after
`cessation of tamoxifen and norethisterone acetate with dura-
`tions of response similar to those found with additive ther-
`apy. Assessment of WR may represent a way of prolonging
`the overall
`response duration in patients with relatively
`indolent metastatic breast cancer, particularly in soft tissues.
`
`Key words: breast cancer, tamoxifen, withdrawal response
`(WR)
`
`Introduction
`
`‘rebound regressions’ of
`responses or
`Withdrawal
`breast cancer werefirst described in 1949 in a group of
`patients who had initially responded to androgens.
`When the tumoreventually progressed on androgen
`therapy a further response was obtained whentreat-
`ment was stopped [1]. WRs were later reported after
`cessation of therapy with estrogensin patients who had
`initially responded [2]. WRs in patients who failed to
`respondto first line treatment with estrogens were first
`described in 1956 [3]. In these early studies all WRs
`were seen after treatment with either estrogens or
`androgens and none were reported after cessation of
`corticosteroids or the progestogensin use at that time
`[4].
`Thefirst reported WR after cessation of tamoxifen
`was seen in a patient with parenchymal lung metastases
`[5]. There was no evidence of an initial response to
`tamoxifen in this patient, indeed the growth rate of the
`metastases appeared to be stimulated by treatment.
`Withdrawal of tamoxifen resulted in a partial response
`whichlasted for at least six months. However,clinicians
`doubted that WRs were observable after tamoxifen
`since, in another study, no WRs were seen in 19 patients
`[6]. Since that time there have been twocase reports [7,
`8] and two series [9, 10] indicating that WRs do occur
`after stopping tamoxifen therapy. Onestudy is particu-
`larly important since it gives some indication of the
`
`incidence of WR after tamoxifen [9]. Sixty-one con-
`secutive patients were assessed for WR and there were
`six (9.8%) responses.
`We began to assess WR in ourclinic in 1980 and
`now report several new examples of WR after initial
`response or no response to tamoxifen. There have been
`no previous reports of WR after cessation of second
`endocrine therapy and we present evidence for this
`phenomenon and also evidence that WR can occur
`after cessation of progestogens.
`
`Patients and methods
`
`Patients
`
`Thefirst evaluation of WR was in January 1980 and the final one
`included in this analysis was in December 1988. Seven patients were
`assessed for WR after progression on adjuvant tamoxifen, 72 after
`failure of first endocrine therapy for advanced disease and 21 after
`failure of second orthird line therapy for advanced disease. Patients
`were selected for assessment of WR because they had relatively
`indolent disease and it was thought unethical to assessall patients in
`this way. Therefore during the period of study a further 194 patients
`were treated with immediate second line additive endocrine therapy
`and 128 were treated immediately with chemotherapy after failure
`of first endocrine therapy.
`
`Pretreatment characteristics
`
`The characteristics of patients assessed for WR after failure of first
`line endocrine therapy are shown in Table 1. Compared with patients
`
`AstraZeneca Exhibit 2016 p. 1
`InnoPharma Licensing LLC v. AstraZeneca AB IPR2017-00904
`Fresenius-Kabi USA LLC v. AstraZeneca AB IPR2017-01910
`
`
`
`612
`
`Table J. Characteristics of patients at the time of second therapy.
`
`Table 2. Responseto first endocrine therapyinall patients.
`
`Treatment
`
`Withdrawal Additive
`(t)
`endocnne
`(2)
`
`p
`(1 vs.2)
`
`p(1+2
`vs. 3)
`
`Second therapy
`
`Withdrawal Additive
`(1)
`endocrine
`(2)
`
`p
`(1 vs.2)
`
`12 (8)
`37 (24)
`
`31 (20)
`
`72 (48)
`
`21 (34)
`
`13 (21)
`
`194
`72
`Patient numbers
`Response to 1st endocrine therapy
`Complete (CR)
`8 (13)
`Partial (PR)
`20 (32)
`No change
`(NC)
`Progression
`(PD)
`Nat evaluable
`(NE)
`Time in months
`Surv. from
`presentation
`Disease free
`Interval
`Time to progres-
`sion (on Ist
`therapy)
`Survival from
`Ist therapy
`Duration of
`response
`
`10
`
`73
`
`31
`
`43
`
`44
`
`15
`
`42
`
`63
`
`28
`
`8
`
`31
`
`14
`
` p (1+2
`vs. 3)
`
`Chemo-
`therapy
`(3)
`
`128
`
`0 (0)
`8 (8)
`
`0.004
`
`13 (12) <0.001
`
`85 (80)
`
`22
`
`42
`
`25
`
`3
`
`19
`
`11
`
`NS
`
`NS
`
`0.006
`
`0.002
`
`NS
`
`<0.0001
`
`0.05
`
`<0,0001
`
`<0.0001
`
`<0.0001
`
`72
`
`63
`(32-90)
`
`6 (8)
`16 (22)
`
`Patient numbers
`Median age at
`presentation
`(range)
`Histology
`Infiltrating duct 50 (70)°
`Infiltrating
`lobular
`Other types
`Receptors
`47 (82)
`ER+ve
`10 (18)
`—ve
`15
`NK
`37 (66)
`PRt+ve
`19 (34)
`—ve
`16
`NK
`Dominant site of disease
`Locally
`advanced
`Recurr. soft
`26 (36)
`tissue
`17 (24)
`Bone
`15 (21)
`Lung
`2 (3)
`Liver
`-
`Brain
`Numberofsites of disease
`1
`32 (44)
`2
`25 (35)
`3
`12 (17)
`3+
`3 (4)
`
`12 (17)
`
`194
`
`60
`(25-87)
`
`138 (71)
`
`14(7)
`42 (22)
`
`108 (79)
`29 (21)
`57
`78 (58)
`57 (42)
`59
`
`25 (13)
`
`34 (17)
`70 (36)
`47 (24)
`18 (9)
`-
`
`76 (39)
`70 (36)
`37 (19)
`11 (6)
`
`NS
`
`NS
`
`NS
`
`NS
`
`0.007
`
`NS
`
`Chemo-
`therapy
`(3)
`
`128
`
`53
`(25-84)
`
`NS
`
`99 (77)
`
`8 (6)
`21 (16)
`
`NS
`
`42 (45)
`52 (55) <0.0001
`34
`32 (34)
`62 (66) <0.0001
`34
`
`6 (5)
`
`23 (18)
`27 (21) <0.0001
`49 (38)
`21 (16)
`2 (2)
`
`31 (24)
`55 (43)
`22 (17)
`20 (16)
`
`0.0007
`
`rank test [14]. The Chi-square test was used to compare tumorre-
`sponse categories.
`
`Other methods
`
`(%).
`"
`NS = notsignificant.
`
`Estrogen (ER) and progesterone receptors (PR) were assayed as
`previously described [15].
`
`who were treated with additive second line endocrine therapy they
`were significantly more likely to have soft
`tissue disease only.
`Patients given chemotherapy as second treatment weresignificantly
`more likely to be receptor negative, to have liver and lung as domi-
`nant sites of disease and to have multiple sites of disease compared
`with the endocrine treated groups. The response to first endocrine
`therapy in the three groups is shown in Table 2. The WR group had
`a significantly higher response rate (79%: CR + PR + NC) than
`those treated later with second line additive endocrine therapy
`(52%) or with chemotherapy (20%). The WR group also had a sig-
`nificantly longer time to progression and survival from first therapy
`than the other two groups.
`
`Assessmentof response
`
`The criteria used for assessment for tumor responses were those
`defined by Haywardet al. for the UICC[11]. Patients were examined
`monthly after withdrawal of therapy. Lung metastases were assessed
`by monthly chest x-rays and bone changes at two monthly intervals.
`The duration of response was taken from the start of treatment to
`the date of objective tumor progression. For WR this time was taken
`from the cessation of tamoxifen to the time of progression; to be
`defined as ‘no change’ (NC) the disease had to be stable for a period
`of at least six months. We have previously shownthat patients who
`have NC for at least six months have the same survival and steroid
`hormonereceptor characteristics as those with a PR |12]. For these
`reasons we believe that NC for 6 monthsis a ‘response’.
`
`Statistical methods
`
`Time to progression and survival curves were calculated according
`to the method of Kaplan and Meier [13] and comparedusing the log
`
`Results
`
`Withdrawalresponseafter adjuvant tamoxifen
`
`One of 7 patients had a WR when adjuvant tamoxifen
`therapy was stopped at the timeof relapse. All relapses
`were in soft tissues. The responder was a 70-year-old
`woman who relapsed on the chest wall and axillary
`lymph nodes 11 months after mastectomy; she had NC
`for 10 months and then progressed again; tumorster-
`oid receptorstatus was not known.
`
`Withdrawal responseafter first additive endocrine ther-
`apy for advanced breast cancer
`
`Seventy-two patients were assessed for a WR after pro-
`gressing on first line endocrine therapy for advanced
`disease. Of these, 65 were evaluable for response. Non-
`evaluability was due to the presence of sclerotic bone
`metastases or
`insufficient observations made. Five
`(8%) of the assessable patients had a partial response
`(PR) for 28+, 22, 8, 6 and 3 months, respectively, while
`a further 14 (21%) showed no change (NC) in tumor
`size for a period of more than six months. The median
`duration of PR was 8 months and for NC the median
`was 11 months.
`Thefirst endocrine therapyin ail patients was either
`tamoxifen alone (64), tamoxifen in combination with
`
`AstraZeneca Exhibit 2016 p. 2
`
`
`
`613
`
`In 9 patients the duration of the WR was longer than
`the duration of the responseto first therapy and in 9
`the duration of the first response was longer. In one pa-
`tient the duration of first response was not clear be-
`cause of missing evaluations. There was no significant
`correlation between the duration of response to first
`therapy and duration of the WR.
`
`Comparison of WR with response to immediate second
`additive endocrine therapy afterfailure offirst endocrine
`therapy
`
`prednisolone (4) or with medroxyprogesterone acetate
`(3); one patient was assessed after withdrawal of
`megestrol acetate. One WR wasseenafter cessation of
`tamoxifen and medroxyprogesterone acetate (NC); the
`remainder of WRs were seen after cessation of tamo-
`xifen treatment alone. The details of all the WRs seen
`after first endocrine therapy are shown in Table 3. Re-
`sponses were mainly seen in advanced primary tumors
`and in skin or nodal metastases. However there were 2
`responses in parenchymal lung metastases and 2 in bone
`metastases. ER status of the primary tumor was known
`in 17 responders and was positive in 15 (88%). PR
`status was known in 17 and waspositive in 12 (71%).
`
`Table 3. Characteristics of responders to withdrawalof first endo-
`crine therapy.
`
`Age Site(s) of
`(yts) disease
`
`PR
`ER
`(f.mols/
`mng cyt.
`protein)
`
`Re-
`sponse
`to Ist
`ther.
`
`Dura- Response Dura-
`tion
`to WR
`tion
`(mths)
`(mths)
`
`During the period that the above 72 patients were as-
`sessed for WR after progression on first endocrine
`therapy for advanced disease, a further 196 patients
`were treated with second additive endocrine therapy
`immediately after progression on first line treatment.
`There was nosignificant difference between the ER
`and PR receptor distribution of the tumors from the
`two groupsbut the second additive therapy group were
`significantly more likely to have distant metastases
`
`
`77~Lung parenchyma NK NK_~PR 32 PR 28+
`
`
`54
`ST, bone (NE),
`38
`300)
`NC
`12
`PR
`22
`(Table 1). Although the two groups are not strictly
`PE
`comparable because theyare selected, it is of interest
`O8
`PR
`2
`PD
`242
`12.
`ST, PE
`47
`6
`PR
`8
`NC
`Q
`79
`~ST
`79
`that the overall response to second additive endocrine
`
`
`69 Adv. pnmary+ST=12 Q PR 35 PR 3
`
`
`
`
`therapy was significantly greater in the group not as-
`71
`Lung parenchyma
`118
`231
`PR
`57
`NC
`40
`sessed for WR (response rates 30% vs. 35% [CR + PR
`74
`ST
`89
`9
`CR
`27
`NC
`32
`56
`ST, bone (NE)
`0
`67
`NE
`14
`NC
`22
`+ NC], p= 0.03 (Table 4). No CRs were seen after
`
`
`86 Adv. primary+ST=.23 26 NC 8 NC 20
`
`
`
`
`assessment of WR but there were 6 (4%) with additive
`67
`ST
`50
`10)
`NC
`22
`NC
`15
`second endocrine therapy. However, there were no sig-
`77
`Adv. pnmary
`33
`16
`PR
`16
`NC
`14
`nificant differences in time to progression or duration
`75
`Adv. primary
`20
`19
`NC
`7
`NC
`12
`85
`Adv. pnmary+ST
`8&8
`670
`NC
`9
`NC
`11
`of response between the two groups. The survival from
`58
`Bone, PE
`S85
`80
`CR
`27
`NC
`10
`the start of second therapy wassignificantly poorer in
`
`
`
`
`47 NK_NEST, bone NK - NC O8
`
`
`56
`ST
`0
`0
`PD
`4
`NC
`O8
`the additive group, which is presumably a reflection of
`61
`ST
`190
`93
`CR
`55
`NC
`07
`their more extensive disease.
`
`
`57. Adv. pnmary+ST=21 QO PD 4 NC 07
`
`
`
`
`75
`ST, bone (NE)
`217
`QO
`NE
`19
`NC
`06
`
`response; NC=No change for
`CR= Complete response; PR= Partial
`>6/12; PD = Progressive disease; ST = skin or node recurrences; PE =
`Pleural effusion; NE = Not evaluable; NK = Not known.
`
`Relationship ofresponse to WR to responseto first endo-
`crine therapy
`
`WRs wereseenafter all types of response tofirst ther-
`apy. Eight complete respondersto first therapy were
`assessed for WR; of whom 5 were evaluable: 3 of the
`evaluable patients had NC for 32, 10, and 7 months,
`respectively. Eighteen of 20 partial responders to first
`therapy were evaluable; there were two PRs of 28+ and
`3 months, and 2 NC of 40 and 14 months. Twenty of
`21 patients with NC on first therapy were evaluable;
`there were 2 PRs of 22 and 6 months and 4 NCof 20,
`15, 12, and 11 months duration. All 13 patients with
`PD on first therapy were evaluable; there was one PR
`of 8 months and 2 NC of 8 and 6 months, respectively.
`Three NC withdrawal responses of 22, 8 and 6 months
`duration were seen in 10 patients not evaluable forfirst
`endocrine therapy. The proportion of WR in each
`group were notsignificantly different.
`
`Table 4. Response rate and response duration to WR, compared
`with second additive endocrine therapy.
`
`Second therapy
`With-
`(%)
`Additive
`(%)
` p
`drawal
`endocrine
`
`
`n
`
`72
`
`194
`
`CR
`PR
`NC
`PD
`NE
`Surv. from 2nd
`therapy (mths)
`Time to progression
`(mths)
`Duration of response
`
`10 10(mths) NS
`
`0
`5
`14
`46
`7
`
`25
`
`3
`
`
`
`(0)
`(8)
`(22)
`(70)
`
`6
`26
`17
`91
`54
`
`16
`
`3
`
`
`
`(4)
`(19)
`(12)
`(65)
`
`0.03
`
`0.004
`
`NS
`
`Endocrine therapy after assessment of WR
`
`Fifty-three of the 72 patients assessed for WR went on
`to have further endocrine therapy. Forty-two of the 53
`were assessable for response. The reasonsfor failure to
`give further additive therapy were death of the patient
`(n= 9),
`treatment with chemotherapy (n =9) and a
`
`AstraZeneca Exhibit 2016 p. 3
`
`
`
`614
`
`continuing WR onepatient with lung metastases after
`28+ months of follow up. Eighteen (42%) of patients
`responded to second additive endocrine therapy (6PR
`and 12NC). Ten patients who had a WR wereassess-
`able and eight (80%) responded to further endocrine
`therapy. Six of these 8 had responded tofirst additive
`endocrine therapy. Thirty-one patients did not have a
`WR and of these 10 (32%) responded to further endo-
`crine therapy. Nine of these 10 had respondedtofirst
`additive therapy. The median duration of response to
`third endocrine therapy was 12 months. A flow chart of
`the outcomeofall of this group of patients is shown in
`Fig. 1.
`
`EFFECT OF
`WITHDRAWAL
`(n= 65)
`
`oP
`SECOND ADDITIVE
`(el)
`
`NOT TREATED
`OR EVALUABLE POR
`ND ENDOCRINE
`
`RESPONSE
`TO FIRST
`
`NO RESPONSE
`TO FIRST
`
`(°)
`
`EVALUABLE
`FOR FIRST
`THERAPY
`
`®
`@
`
`@)
`
`~)
`
`O
`
`(7)
`
`(9)
`©
`@)
`@)
`
`©
`@)
`
`(3)
`
`()
`
`(*)
`CO)
`
`7
`
`6
`
`1
`
`3
`
`6
`
`Fig. 1. Overview of responsestofirst additive therapy, to withdrawal
`of tamoxifen, and to second additive therapy. Upper figures =
`O responders; Lowerfigures = © progressors.
`
`There was a small group of 6 patients who were
`assessable forall three therapies and who responded to
`all three. Two of these are of particular interest since
`they had a second response to tamoxifen after a period
`of no treatment during which a WR was assessed. Both
`had slowly progressive chest wall disease which re-
`sponded completely to the first treatment with tamo-
`xifen. After progression and cessation of tamoxifen,
`WRs were seen (NC) for 11 and 20 months, respec-
`tively; both patients then went on to have anotherre-
`sponse to tamoxifen when it was reintroduced at the
`same dosage of 20 mg per day but did no respond to a
`second withdrawal of tamoxifen.
`
`Withdrawal responses after second or third line endo-
`crine therapy
`
`Eighteen patients were assessed for WRs after cessation
`of second therapy and 3 after withdrawal of third ther-
`apy. There were 4 NC (22%) for 14, 14, 9, and 7
`months after withdrawal of second line additive endo-
`crine therapy but no responses were seen after with-
`drawal of third line endocrine therapy. Three stabilisa-
`tions were seen after withdrawal of norethisterone
`
`acetate. The durations of responses after cessation of
`norethisterone acetate were 14, 14, and 7 months,re-
`spectively. A fourth patient was treated with 20 mg of
`tamoxifen per day as first therapy and had a complete
`response for 12 months; the dose was increased to 40
`mg per day at progression and shehada further period
`of no change for 26 months. Tamoxifen was then with-
`drawn and she had a WR of 9 months duration.
`
`Discussion
`
`This study confirms that WRs are detectable after cessa-
`tion of additive therapy with tamoxifen used as first line
`endocrine therapy for advanced breast cancer
`[5,
`6-10]. There is only a single previous report of WR
`after no initial response to tamoxifen [5] and we report
`3 further examples. In addition we report for the first
`time WR after norethisterone acetate and after cessa-
`tion of second line endocrine therapy.
`Five of the 24 WRs reported here werepartial remis-
`sions and the remainder were in the ‘no change’ cate-
`gory. Although ‘no change’ is regarded byinternational
`criteria as a ‘response’ [11] a question remains as to
`whether this is justifiable. Since short periods of NC
`maybe difficult to distinguish from slowly progressive
`disease we have taken a minimum period of six months
`of NC for a patient to qualify for this category of re-
`sponse. In previous studies we and others [{12, 16, 17]
`have shown that, if this period is taken, the patients
`with NC have similar durations of response and sur-
`vival to those with an objective PR and their tumours
`have similar ER and PR positivity [12]. All 19 NC had
`signs of objective progression on additive therapy be-
`fore it was stopped and the period of NC lasted from 6
`to 40 months.
`
`Incidence of withdrawal response
`
`Becauseof patient selection for assessment of WR this
`study gives only a broad indication of the true inci-
`dence of WR in a population of womenfailing first line
`endocrine therapy for advanced disease. In this regard
`the paper by Canneyetal. [9] is importantsinceit is the
`only one, to our knowledge, where consecutive patients
`were assessed. They detected 6 responses (9.8%; 1 CR,
`4PRs and | NC) in 61 patients studied; all responses
`were in soft tissue disease and occurred in patients who
`had previously responded to first endocrine therapy.
`This figure is lower than the overall WR rate after first
`endocrine therapy of 30% reported here. However our
`rate falls to 6.2% (19/308) when the total numberof
`evaluable patients forall types of second therapyin the
`whole study population is considered as the denomina-
`tor. It is possible that this figure would be higherifall
`of our patients, particularly those treated with second
`line additive endocrine therapy, were assessed for WR.
`In the earlier reports of WRs following cessation of
`treatments with estrogens and androgens WRs werere-
`
`AstraZeneca Exhibit 2016 p. 4
`
`
`
`Table 5. Withdrawal responses after first endocrine therapy.
`
`Additive
`(reatment
`
`WR Total
`assessed
`
`{%o) Total study
`(%o)
`Reference
`
`population
`
`615
`
`line endocrine therapy suggesting that there is a small
`group of highly endocrine responsive tumours which
`will not respond to endocrine withdrawal. Conversely
`there is a small group of patients who appear to be
`highly responsive in that they responded to withdrawal
`Androgens (64) Notknown=—7°11 Escher 1949 [1]
`
`
`
`Estrogens
`9
`14
`(64) 100
`(9.0) Huseby 1954 [2]
`as well as to two additive therapies (Fig. 1).
`Andr. & estr,
`15°
`8S
`(17) 674
`(2.2)' Kaufman 1961 [4]
`The duration of WR forall patients in this series
`Estrogens
`7
`22
`(32)
`97
`(7.2) Baker 1972 [19]
`ranged from 3-40 months. A similar wide range of
`Estrogens
`&
`32
`(25)
`83
`(9.6) Nesto 1976 [18]
`Tamoxifen
`6
`61
`(0)
`61
`(9.8) Canney 1987 [9]
`durations was reported after WR to androgen and
`Tamoxifen
`19
`65
`(29) 308
`(6.2) This study
`estrogen therapy. Kaufman and Escher[4] pointed out
`that the average duration of response to withdrawal is
`similar to that seen in respondersto additive therapy.
`The same appears to be the case for tamoxifen; the
`median duration of response to withdrawal of tamo-
`xifen in this study was 10 months and was identical to
`the duration of response in the group of patients given
`immediate additive therapy as secondline treatment.
`WRs occur in patients who have not responded to
`first endocrine therapy. This phenomenonwas report-
`ed first by Kaufman and Escher[4]. They saw 6 WRs
`after failure of patients to respond to estrogens and 4
`after failure to respond to androgens: (no denominator
`was given). Baker and Vaitkevicius [19] noted WRs in
`2 of 11 (18%) non-responders to estrogens. Legault-
`Poissonetal. [5] were the first to report WRs after fail-
`ure to respondtofirst line therapy with tamoxifen in a
`patient with lung metastases. The basal doubling time
`of her lung metastases as measured onserial chest
`x-rays was 120 days. After starting tamoxifen there was
`tumor growth stimulation with a reduction in the
`doubling time to 52 days. Treatment was discontinued
`after 110 days therapy and the patient remainedoff all
`therapy; a PR occurred which lasted for more than 6
`months. In the study reported here we found 1 PR and
`2 NC in 13 evaluable patients for WR who had objec-
`tive progression onfirst linc therapy with tamoxifen.
`WRs after stopping progestogens have notbeenpre-
`viously documented. Kaufman and Escher[4] reported
`no responses after 71 trials of withdrawal to unspe-
`cified progestogens. Here we report three WRs after
`stopping norethisterone acetate given as a second line
`therapy.
`
`‘
`
`Includes additional WR seen after no response tofirst therapy.
`
`ported only in a subset of the total study population
`(Table 5). For example Huseby [2] reported WR after
`cessation of estrogens in 9 of 14 initial responders to
`estrogen but did not report assessment of WR in the
`remaining 86 patients in the total study population of
`100 patients. When these are included in the analysis
`the overall rate of WR is 9%. We show in table 9 the
`proportion of WR in relation to total patient numbers
`in all the reported series. The overall absolute inci-
`dence of WR ranges from 3 to 10%.It is probable that
`3% is an underestimate due to failure to assess all
`potential responders for withdrawal of additive endo-
`crine therapy. As judged by the consecutive series of
`Canneyetal. [9] the figure may be nearer 10% of the
`total number of patients whofail first line endocrine
`therapy.
`
`Sites ofresponse
`
`Although WRs maybeof value in approximately 10%
`of patients they represent a subgroup whotend to have
`soft tissue disease. This observation also applies to the
`old studies with estrogens and androgensas well as the
`more recent studies with tamoxifen. However, there
`are exceptions to these observations. At least 6 WRs
`after tamoxifen have been reported in lung metastases
`[5, 7, 8, 10] (and 2 in this study) and WRsafter andro-
`gens and estrogens have been reported in bone [3, 18],
`liver [18] and brain secondaries[4].
`
`Comparison of WR with response to secondline additive
`endocrine therapy
`
`Mechanism of withdrawal response
`
`It is well documented that a proportion of patients will
`have a response to a second additive endocrine ther-
`apy. In the group of patients who had a second additive
`therapy in this study 35% responded compared to the
`30%response rate in the potentially more favourable
`group who were assessed for a withdrawal response.
`Despite there being CRs and more PRsin the additive
`group the median durations of response and the times
`to progression of the two groups were notsignificantly
`different. It has to be emphasised, however, that the
`two groups are not strictly comparable because those
`assessed for WR werehighly selected. However eight
`of the 10 patients who responded to additive therapy
`after failing to have a WR had also responded to first
`
`The early studies with androgens and estrogens and
`the morerecentstudies with antiestrogens and proges-
`togens indicate that WR is a definite phenomenon
`which occurs in a subgroup of prognostically favour-
`able patients. It appears that similar proportions of pa-
`tients with similar clinical characteristics respond to
`withdrawal of each type of therapy.
`It is tempting to suggest that because withdrawal of
`treatment results in a change from tumor growth to
`tumor regression or stabilisation, that under certain
`circumstances additive endocrine therapy may stimu-
`late tumor growth. Removal of the growth stimulus by
`stopping treatment then results in growth inhibition
`because of absenceof a ‘trophic’ hormone.
`
`AstraZeneca Exhibit 2016 p. 5
`
`
`
`616
`
`low concentrations (1077-10 M)and in the
`At
`absence of estradiol, tamoxifen stimulates the prolif-
`eration of several estrogen receptorpositive cell lines in
`vitro [20-22]. Furthermore low dose stimulation may
`be responsible for the tumor flare phenomenon re-
`ported when tamoxifen serum levels are gradually ris-
`ing duringthefirst weeks of therapy [23]. Stimulation of
`proliferation has also been demonstrated at therapeutic
`levels (10M) of tamoxifen. Tumor cells from the
`pleural effusions of two patients who had failed tamo-
`xifen therapy, were stimulated to proliferate in vitro in
`the presence of 10°°M tamoxifen [24]. These two pa-
`tients later went on to have a clinical WR when tamo-
`xifen therapy was stopped. Stimulation of colony for-
`mation of primary human mammary tumorcells by
`10°M tamoxifen in the soft agar clonogenic assay has
`been demonstrated [25, 26]. In one study [26] nearly
`30% of specimens had a > 100%survival of colonies in
`tumors exposed to tamoxifen and also to therapeutic
`concentrations of medroxyprogesteroneacetate.
`A phenomenonsimilar to a WR in humans can be
`demonstrated in nude mice. Tamoxifen inhibits the
`proliferation of the human mammary tumorcell line
`MCEF-7 when implanted subcutaneously in nude mice.
`However, after prolonged treatment tumors regrow and
`can be shown to have dependance upon tamoxifen;
`growth ceases when either
`tamoxifen treatment
`is
`stopped or the mice are treated with the pure antiestro-
`gen ICI 164 384 as well as tamoxifen [27]. Presumably,
`unlike tamoxifen, the latter compound prevents binding
`of the antiestrogen receptor complex to DNA andthus
`inhibits its agonist activity [28]. Agonist activity at
`therapeutic concentrations of tamoxifen is also some-
`times demonstrable in the normal endometrium and
`myometrium in patients undergoing treatmentfor their
`breast cancer [29, 30]. In addition an endometrial can-
`cer cell line grown in nude mice can be shownto prolif-
`erate more rapidly when treated with tamoxifen whilst
`MCEF-7 cells implanted into the opposite flank are in-
`hibited [31]. The latter experiment suggests, although
`docs not prove, that alterations in tamoxifen metabo-
`lism are unlikely to account for tumorstimulation. In
`patients, serum levels and the proportion of various
`tamoxifen metabolites show only minor changes over
`prolonged periods [32, 33]. However, changes in the
`intracellular metabolism of tamoxifen which have been
`described recently may be important[34].
`A satisfactory hypothesis to explain WR must ac-
`count for the apparent change of a drug from antago-
`nist to agonist during therapy, how removal of the puta-
`tive agonist can result in prolonged WR, why such an
`effect can be seen using several different classes of
`endocrine agent and whyreintroduction of the same
`initial additive endocrine therapy at progressionafter a
`WRcan result in a second response.
`Onepossible explanation is that some tumors have
`several clones of receptorpositive cells each with a dif-
`ferent sensitivity to a particular hormone agonist or
`antagonist. A fundamental feature of tumorcells ap-
`
`pears to be their heterogeneity [35]; examination of a
`wide range of tumorcharacteristics showsthat a major-
`ity are variable. Similar observations apply to normal
`tissues such as breast lobules. Wide variation in sensi-
`tivity to endogenous hormonesis seen in early preg-
`nancy, where highly proliferative or secretory lobules
`are seen adjacent to quiescent ones, as judged bylight
`microscopy, suggesting marked heterogeneity of hor-
`mone responsiveness between lobules. Clones of
`human mammary cell lines with widely different sensi-
`tivities to hormoneshavealso been reported [36].
`It is possible that a balance between multiple clones
`within a tumoris altered by the selective pressure of
`endocrine therapy. Clones inhibited by an agent, may
`regress whilst those stimulated to grow might gradually
`become dominant. Removal of the selective pressure
`could allow the previous ‘balanced’ interactions be-
`tween clones to reoccur: under these circumstances a
`second response to the same agent may occur as was
`demonstrable in two of our patients. The clonal hy-
`pothesis could be tested directly by extending the
`experiments of Simonetal. [24] by exposing progres-
`sing tumours to pharmacological concentrations of
`endocrine agents in vitro. In the clinic it could be tested
`by treatment with pure antiestrogens which have been
`shown in vitro and in animals to be devoid of agonist
`activity. Treatment with the pure antiestrogen ICI
`182780 [37], which is currently entering clinical trial,
`may possibly increase response rates and response
`durations to therapy by obviating the apparent agonist
`effects of most of the commonly used additive endo-
`crine therapies.
`
`Acknowledgement
`
`Mrs Linda Ashcroft for data management. Supported
`by The Cancer Research Campaign.
`
`References
`
`improvement of inoperable breast car-
`1. Escher GC. Clinical
`cinomaundersteroid treatment. In Proceedings of the 1st Con-
`ference on Steroid Hormones and Mammary Carcinoma.
`Chicago. The Therapeutic Trials Committee of the Council on
`Pharmacy and Chemistry of the American Medical Assn 1949;
`92-9.
`
`2. Huseby RA.Estrogen therapy in the managementof advanced
`breast carcinoma. Am J Surg 1954; 20: 112.
`3. Lewison EF, Trimble FH, Ganelin RS. Advanced mammary
`cancertreated with sex hormones. J Am Med Assn 1956; 162:
`1429-37.
`
`4. Kaufman RJ, Escher GC. Rebound regression in advanced
`mammary carcinoma. Surgery, Gynecology & Obstetrics 1961;
`635-40.
`
`5. Legault-Poisson S, Jolivet J, Poisson R et al. Tamoxifen-in-
`duced tumor stimulation and withdrawal response. Cancer
`Treat Rep 1979; 63: 1839-41.
`6. Beex LVAM, Pieters GFFM, Smals AGH et al. Diethylstil-
`bestrol versus tamoxifen in advanced breast cancer. N Engl J
`Med 1981; 304: 1041.
`7. Belani CP, Pearl P, Whitley NO. Tamoxifen withdrawal
`sponse. Arch Int Med 1989, 149: 449-50.
`
`re-
`
`AstraZeneca Exhibit 2016 p. 6
`
`
`
`Stein W, Hortobagyi GN, Blumenschein GR. Response of
`metastatic breast cancer to tamoxifen withdrawal: a report of a
`case. J Surg Oncol 1983; 22: 45-6.
`. Canney PA,Griffiths T, Latief TN et al. Clinical significance of
`tamoxifen withdrawal response. Lancet 1987; 1: 36.
`OsborneCK,von Hoff DD, Mullins K. Endocrine therapytest-
`ing of human breast cancers in the soft agar clonogenic assay.
`Rudolph R. Response of metastatic breast cancer to tamoxifen
`Br Cancer Res Treat 1985; 6: 229-35,
`(TAM) withdrawal. Proc Am Soc Oncol 1986; 5: 270.
`Hayward JL, Carbone PP, Heuson JC et al. Assessmentof re-
`Gottardis MM, Jiang S-Y, Jeng M-Hetal. Inhibition of tamo-
`sponse to therapy in advanced breast cancer. Eur J Cancer
`xifen-stimulated growth of an MCF-7 tumour variant
`in
`1977; 13: 89-94.
`athymic mice by novel steroidal antiestrogens. Cancer Res
`1989; 49: 4090-3.
`Howell A, Mackintosh J, Jones M. The definition of the ‘no
`Fawell SE, White R, HoareSet al. Inhibition of estrogen recep-
`change’category in patients treated with endocrine therapy and
`chemotherapy for advanced carcinoma of the breast. Eur J
`tor-DNA binding by the ‘pure’ antiestrogen ICI 164 384 ap-
`CancerClin Oncol 1988; 24: 1567-72.
`pears to be mediated by impaired receptor dimerization. Proc
`Natl Acad Sci 1990; 87: 6883-7.
`Kaplan EL, Meier P. Nonparametric estimation from incom-
`plete observations. J Am Stat Assoc 1958; 53: 457-81.
`Boudouris O, Ferrand S, Guillet JL et al. Effets paradoxaux du
`tamoxifene sur l'utérus de la femme. J Gynecol Obstet Biol
`Pet