p21waf1 expression and endocrine response in breast cancer

J. Pathol. 188: 126–132 (1999)

p21WAF1 EXPRESSION AND ENDOCRINE RESPONSE INBREAST CANCER

. 1, . . 1, ’1, . 2, . . 3, . 4 . 1*

1Tenovus Cancer Research Centre, University of Wales College of Medicine, Heath Park, Cardiff CF4 4XX, U.K.2Imperial Cancer Research Fund, Clinical Oncology Unit, Guy’s Hospital, London SE1 9RT, U.K.

3Department of Surgery, City Hospital, Hucknell Road, Nottingham, U.K.4Department of Histopathology, City Hospital, Hucknell Road, Nottingham, U.K.

SUMMARY

An immunocytochemical assay for the p53-regulated protein product of the WAF1/Cip1 gene, p21WAF1 (p21), was developed andapplied to archival primary breast tumour material from 91 patients whose subsequent recurrent disease was treated with assessablecourses of endocrine therapy. Nuclear localization of p21 protein was observed in 76 (82·4 per cent) cases. Status cut-offs wereestablished and 29 (31·9 per cent) were deemed negative, 39 (42·9 per cent) weakly positive, and 23 (25·3 per cent) strongly positive. p21status was inversely correlated with p53 protein (p=0·047) but did not relate to oestrogen receptor (ER) status, response to endocrinetherapy, or time to further disease progression (TTP). Highly p21-positive patients had a significantly improved overall survival time(p=0·020). Co-assessment of p21 and p53 subgroups revealed p21+/p53" patients to have good survival characteristics, whilstp21"/p53+ patients did poorly (p=0·008). The p21"/p53" patients overall did intermediately well, but Ki67-defined cellularproliferation analysis of these revealed two subclasses: those with high proliferation and poor survival times resembling the p21"/p53+phenotype, and those with less proliferative tumours with good survival, similar to the p21+/p53" group. The significance of theseresults is discussed in the light of recent research concerning the role of p21 and p53 in breast cancer aetiology. Copyright ? 1999 JohnWiley & Sons, Ltd.

KEY WORDS—p53; tumour suppressor genes; breast cancer; p21; endocrine response; prognosis

*Correspondence to: Dr R. I. Nicholson, Breast CancerLaboratory, Tenovus Cancer Research Centre, University of WalesCollege of Medicine, Heath Park, Cardiff CF4 4XX, U.K.

INTRODUCTION

Wild-type p53 protein1 plays an important role inmaintaining the integrity of the human genome,2 alert-ing the cell to DNA damage3 and preventing the com-promised cell from proliferating until conditionsimprove. If repairs cannot be effected, p53 promotes theremoval of the defective cell via apoptosis4 and thusensures that defective genetic material is not passedto subsequent generations of cells. This tumour-suppressing role is partly mediated by the transcriptionalactivation of a number of p53-response element(p53-RE)-containing genes such as WAF1/Cip1,5,6

whose p21 protein product is believed to block cyclin/cyclin-dependent kinase (CDK) complex activity7 andthus prevent the passage of cycling cells from G1 toS phase. p21 is therefore an essential part of thep53-mediated growth arrest pathway elicited by DNAdamage. It is well recognized, however, that p21 can alsobe regulated by p53-independent mechanisms.8–10

Mutations in the p53 gene resulting in a loss of p53function are the most frequently encountered geneticchange in cancer and occur in approximately one-thirdof breast tumours.11 Biologically significant mutationsoften result in the synthesis of a mutant protein capableof binding to the p53-RE of responsive genes, but are

Contract/grant sponsor: Tenovus Organisation.

CCC 0022–3417/99/070126–07$17.50Copyright ? 1999 John Wiley & Sons, Ltd.

incapable of recruiting the machinery necessary foreliciting gene transcription. One significant consequenceof mutant p53 production is its subversive effect onthe p53-RE-bearing MDM212 feed-back loop whichnormally regulates expression from the p53 gene andencourages rapid degradation of p53 protein,13 thusallowing mutant protein levels to accumulate. Thisaccumulation and extended half-life of mutant p53are utilized in immunocytochemical assays (ICAs),wherein observable p53 staining is assumed to be almostexclusively of the mutant protein.

To date, a number of ICA-based studies relating p53and breast cancer aetiology have been performed.14–16

Thus, elevated p53-ICA positivity has been shown toassociate with features of loss of differentiation (e.g.high histological grade, loss of tubule formation,increased nuclear pleomorphism, and mitoticactivity),15,17–19 with markers indicative of poor prog-nosis (e.g. ER negativity, EGFR positivity, and c-erbB2positivity), with shortened disease-free and overallsurvival periods,17,18,20 and also with failure of endo-crine therapy in ER-positive advanced breast cancerpatients.17 However, the role of p53-responsive genes inendocrine-responsive breast cancer remains less wellresolved. Thus, we present an immunocytochemicalanalysis of p21 expression, a downstream marker ofwild-type p53 functionality, and report its significance ina population of endocrine therapy-treated advancedbreast cancers in relation to established biologicalmarkers, clinical response, and survival.

Received 1 April 1998Revised 30 September 1998Accepted 20 January 1999

127p21WAF1 IN BREAST CANCER

MATERIALS AND METHODS

and mutant p53 protein are given elsewhere.

Patients

For inclusion in the study, excised primary carcinomamaterial for histology and immunocytochemistry andthe subsequent availability of a secondary lesion tomonitor the efficacy of endocrine therapy were required.Ninety-one patients presenting to the breast clinics ofProfessor R. Blamey between 1987 and 1993 weredeemed eligible (mean age at primary diagnosis of 55years, range 25–83 years). Of these, 65 (71 per cent) werepost-menopausal. Tumours were predominantly ofgrades II [38 (41·8 per cent)] and III [50 (54·9 per cent)]21

and comprised 52 ductal NOS, 12 lobular, nine lobularvariants, seven tubular mixed, one encysted papillary,two medullary, and eight atypical medullary carci-nomas.22 The secondary tumours comprised 41 (45 percent) locally advanced primary and 50 (55 per cent)metastatic disease cases.

Endocrine therapy

Responses to endocrine therapy were assessed inaccordance with UICC criteria,23 with a minimumduration for response and static disease of 6 months.24

Eighteen pre-menopausal patients received an LH-RH-agonist, goserelin (Zoladex ICI 118630 3·6 mg depot/28 days), alone; six received goserelin with tamoxifen(Nolvadex, ICI46474, 20 mg twice daily); and twotamoxifen alone. Sixty-one post-menopausal patientswere given tamoxifen alone, three received the pro-gestogen megestrol acetate (Megace, Bristol-Myers,160 mg twice daily), and one tamoxifen with goserelin.

Tissue samples

Representative portions of the primary carcinomawere divided and either fixed in neutral buffered for-malin for 24 h and processed into paraffin blocks, orsnap-frozen and stored in liquid nitrogen. Assays forp53 (paraffin) were performed at the Imperial CancerResearch Fund, London and for p21 (paraffin), ER(frozen), and Ki67 (frozen) at the Tenovus CancerResearch Centre, Cardiff.

p21WAF1 immunocytochemistry

Paraffin sections (5 ìm) were cut onto silanized glassslides (TESPA), dried overnight at 37)C, dewaxed, andrehydrated. After immersion in phosphate-bufferedsaline (PBS), endogenous peroxidases were inhibitedusing 3 per cent hydrogen peroxide (aqueous) for 5 min.Slides were washed in PBS and microwave-treated at560 W for 30 min in citrate buffer (pH 6) for antigenretrieval. After cooling and a PBS wash, slides wereincubated with 1 per cent dried milk powder (aqueous,15 min) to block non-specific antigen localization.Excess block was removed and slides were incubatedwith mouse anti-human p21 monoclonal antibody (1/40dilution, room temperature, overnight; Progen DCS-60,Insight Biotechnology, Middlesex, U.K.). Sections were

Copyright ? 1999 John Wiley & Sons, Ltd.

washed in PBS and DPC buffer wash (DPC, Llanberis,U.K.). Biotinylated anti-mouse IgG secondary antibody(Biomen Ltd., Berkshire, U.K.) was added (1/20 in1 per cent BSA/PBS) for 60 min, then washed off withPBS followed by DPC wash solution. Streptavidin/peroxidase tertiary reagent (Biomen Ltd.) was added(1/20 in 1 per cent BSA/PBS) for 60 min. Sections werewashed and staining was revealed with diaminobenzi-dine hydrochloride/H2O2. Slides were counterstainedwith 0·5 per cent aqueous methyl green.

p53 immunochemistry

Details of the p53 assay utilizing the CMI rabbitpolyclonal antiserum which recognizes both wild-type

25

Oestrogen receptor immunochemistry (ER-ICA)

Oestrogen receptor (ER) was assessed on frozensections using the ERICA monoclonal antibody (AbbottLaboratories, North Chicago, U.S.A.) as previouslydescribed.26,27 The system employs a rat anti-human ERH222 antiserum to localize nuclear ER.

Ki67 immunochemistry

The Ki67 antigen was localized as previouslydescribed27,28 using the Ki67 monoclonal antibody(MO722, Dako Laboratories, Glostrup, Denmark) at adilution of 1/100. This antiserum recognizes nuclearantigen expression throughout cell-cycle phases G1to M.

Analysis of staining

Staining was assessed by two personnel and estimateswere recorded of percentages of cells specifically stainedand, where appropriate, of staining intensity. ForERICA, an Hscore (range 0–300) was calculated27

(Table I) and previously established cut-offs for posi-tivity for ER of Hscore>2, shown to predict endocrineresponse effectively, were applied to determine two ERstatus categories.27 For p53 staining, a similar calcu-lation based on the sum of an intensity and % score wasapplied to determine three p53 status groups (detailed inTable I).

Ki67 staining patterns vary depending upon thecell-cycle state of each cell, making staining intensityinterpretations impractical. Consequently, staining isinterpreted on the percentage of tumour cells stainedonly. Three previously described29 cut-offs for Ki67 of0–10 per cent=negative, 11–29 per cent as low positive,and ¢30 per cent as strongly proliferative have beenshown to have prognostic value and are used here(Table I).

p21 staining was evaluated using the same Hscoreprocedure as for ER (i.e. range 0–300). Cut-offsdelineating three status groups were made and theirvalidation is addressed in the results and illustrated inTable I.

J. Pathol. 188: 126–132 (1999)

128 R. A. MCCLELLAND ET AL.

Some analyses of p21 and p53 data were performedusing both three status groupings ("/+/+ +) and,following the combination of negatives and weaklypositives, two status groups ("/+).

Statistics

Correlations between immunocytochemistry statusand response data were determined using the Pearsonchi-square test and Fisher’s exact test using p<0·05 forsignificance. Survival analyses of TTP and overall sur-vival were based on the Kaplan–Meier method with theBreslow procedure for univariate comparison betweengroups (p<0·05 as significant).


Fig. 1—Photomicrograph illustrating specific nuclear p21 immunos-taining of a breast carcinoma

Table I—Cut-off criteria for the immunocytochemical assays

ICAAssessmentmethod Formula Status category 1 Status category 2 Status category 3

ER ER-Hscore (% weakly stained cells#1)+(% moderately stained#2)+(% strongly stained#3)

Hscore: 0<2 negative Hscore: 2¡300 positive N/A

p53 p53-Hscore Sum of [predominant intensitycategory: either 0 (nostaining), 1 (weak), 2(moderate) or 3 (strong)]+[% stained category where:0=0%, 1=>0<25%,2=25<50%, 3=50<75%,4=¢75%]

Hscore: 0–3 negative Hscore: 4–5 weaklypositive

Hscore: 6–7 stronglypositive

Ki67 % positivity Estimation of overall %tumour cells staining

0–10% negative 11–29% weaklypositive

¢30% stronglypositive

p21 ER-Hscore As ER-Hscore above Hscore: 0<3 negative Hscore: 3<30 weaklypositive

Hscore: ¢30 stronglypositive

RESULTS

Immunolocalization of specific p21 protein wasobserved in more than 1 per cent of tumour cells in 75 of91 (82·4 per cent) cases and was almost exclusivelynuclear, with only occasional weak cytoplasmic stainingobserved (Fig. 1). Staining was heterogeneous, with thepercentage of tumour cells stained ranging from 0 to60 per cent (median 5 per cent), but usually staining wasobserved in ¡10 per cent of cells (63/91, 69·2 per cent).Hscore calculations were made and results ranked,enabling the establishment of cut-offs for three apparentgroups (Table I). Thus, tumours with staining levelsbelow quartile Q1 (Hscore 0<3) were deemed negative[i.e. 29 (31·9 per cent)]; levels between Q1 and Q3(Hscore 3<30) were deemed weakly positive [39 (42·9per cent)]; and those above Q3 (Hscore ¢30) wereassumed as strongly positive [23 (25·3 per cent)].

Using the ERICA cut-off criteria, 38 (41·8 per cent)samples were classified as ER-negative and 53 (58·2 percent) as ER-positive.

p53 immunostaining, like ER, was predominantlynuclear and using the cut-offs described, 51 (56 per cent)

tumours were classified negative, 23 (25·3 per cent)weakly positive, and 17 (18·7 per cent) stronglypositive.

Applying the Ki67 cut-offs, 13 (14·3 per cent) tumourswere classified as negative, 17 (18·7 per cent) weaklypositive, and 61 (67 per cent) strongly positive.

To assess the clinical significance of p21 protein, itsexpression was co-analysed with a number of othermarkers. No significant associations were observedbetween p21 status and either histological grade or typeof tumour. The marked predominance of grade II andIII tumours in the study population, however, madeobservation of such an association unlikely.

Using three categories for both p53 and p21 statusrevealed an inverse relationship between them (Pearsonchi square: p=0·047) (Table II). Grouping togethernegatives and weak positives for both markers (TableIII) showed 52/91 (57·1 per cent) samples to be negative/weakly positive for both, whilst only one sample was adouble strongly positive. The inverse correlation was

J. Pathol. 188: 126–132 (1999)


Table II—Comparison of p21 and p53 immunocytochemical staining status using two positive groupsfor each

p53-negativep53-weakly

positivep53-strongly

positive Total

p21-negative 16 (17·6%) 4 (4·4%) 9 (9·9%) 29 (31·9%)p21-weakly positive 23 (25·3%) 9 (9·9%) 7 (7·7%) 39 (42·9%)p21-strongly positive 12 (13·2%) 10 (11·0%) 1 (1·1%) 23 (25·3%)Total 51 (56·0%) 23 (25·3%) 17 (18·7%) 91 (100%)

Pearson chi square p=0·047.

Table III—Comparison of p21 and p53 immunocytochemicalstaining status using combined positive groups for each

p53-negative p53-positive Total

p21-negative 52 (57·1%) 16 (17·6%) 68 (74·7%)p21-positive 22 (24·2%) 1 (1·1%) 23 (25·3%)Total 74 (81·3%) 17 (18·7%) 91 (100%)

Pearson chi square p=0·041. Fisher’s exact test: two-tail p=0·060.

Table IV—Comparison of p21 and ER immunocytochemical staining status using two positive p21 groups

p21-negativep21-weakly

positivep21-strongly

positive Total

ER-negative 17 (18·7%) 14 (15·4%) 7 (7·7%) 38 (41·8%)ER-positive 12 (13·2%) 25 (27·5%) 16 (17·6%) 53 (58·2%)Total 29 (31·9%) 39 (42·9%) 23 (25·3%) 91 (100%)

Pearson chi square p=0·076, NS.

Table V—Comparison of p21 staining status with response to endocrine therapy

Completeresponse

Partialresponse

Staticdisease

Progressivedisease Total

p21-negative 5 5·5%) 2 (2·2%) 3 (3·3%) 19 (20·9%) 29 (31·9%)p21-weakly positive 5 (5·5%) 3 (3·3%) 8 (8·8%) 23 (25·3%) 39 (42·9%)p21-strongly positive 1 (1·1%) 3 (3·3%) 6 (6·6%) 13 (14·3%) 23 (25·3%)Total 11 (12·1%) 8 (8·8%) 17 (18·7%) 55 (60·4%) 91 (100%)


similar (Pearson chi square: p=0·041; Fisher’s exact test,two-tail p=0·060) with 22 (24·2 per cent) tumours p21+/p53" and 16 (17·6 per cent) p21"/p53+.

No significant association was observed between p21and ER status (Pearson chi square: p=0·076) (Table IV),although the highest ER expression was observed in p21positives.

p21 status was further examined in relation to re-sponse to first line endocrine therapy, where responseswere determined on an index lesion of recurrent tumour.No associations were detected using any groupings ofp21 status, nor by analysing response categories individ-ually or following combination of complete and partial


responders with and separately from static diseasepatients (an example is presented in Table V). Furthersubstratification of patients according to their ER statusand repeating these response to therapy analyses also didnot reveal any relationship (data not shown).

Kaplan–Meier survival analysis using the Breslowstatistic to examine equality between groups showed noinfluence of p21 on TTP of patients on endocrinetherapy (p=0·199; strongly positives versus negativesand weakly positives combined).

Overall survival analysis suggested a survivaladvantage of those patients whose tumours were highlyp21-positive when compared with negatives and withweakly positives (Breslow statistic: p=0·055, Fig. 2).Survival curves for the negative and weakly positive dataappear equivalent and distinct from the strongly posi-tives; consequently the analysis was repeated in Fig. 3,where strongly positives were compared with the com-bined negative/weakly positive group and the resultsfound to be significant (p=0·020). Thus, in subsequentsurvival analyses involving p21 and p53, only these twodistinct classifications are used.

p21 survival analysis of ER positive patients onlyshowed no statistically significant relationship (Breslowstatistic: p=0·290).

J. Pathol. 188: 126–132 (1999)


In an attempt better to define the apparent survivaladvantage of patients with highly p21-positive tumours,TTP and survival curves for the groups stratified bycombined p21 and p53 status were performed. As dis-cussed, only a single tumour was highly positive for bothp21 and p53 and it was omitted from these analyses. Nosignificant TTP advantage was afforded to any onegroup (data not shown). Figure 4 illustrates the relevantoverall survival curve obtained for these groups. Thuspatients with p21+/p53" tumours appear to live sig-nificantly longer than p21"/p53+ patients (p=0·012).As illustrated, a large group of patients (n=52) whowere classified as negative for both markers didintermediately well.

As p21 protein is believed to be involved in cell-cyclearrest, it was considered important to analyse Ki67immunostaining in relation to these p21/p53 groupings.Chi-square analysis found no association beween p21and Ki67 using any described cut-offs (Table VI). Figure5 illustrates the percentage of Ki67-stained cells inrelation to the p21/p53 groupings and shows a weak, butnot significant, trend towards decreasing Ki67 withincreased p21 staining (Kruskal–Wallis one-wayANOVA p=0·17). Survival analysis following sub-division of the p21"/p53" subgroup only by Ki67


categories is shown in Fig. 6 and reveals a group ofp21"/p53"/Ki67" patients whose overall survivalappears almost as good as the p21+/p53" phenotype(p=0·055). Analysis of other p21/p53 phenotypes byKi67 was prevented by small subgroup numbers.

Fig. 2—Life-table analysis of the influence of p21 expression on overallsurvival. p21-ICA status subdivided into negative (n=29), weaklypositive (n=39), and strongly positive (n=23). (Breslow statistic:p=0·055)

Fig. 3—Life-table analysis of the influence of p21 expression on overallsurvival. p21-ICA negative category represents both negative andweakly positive cases (n=68). (Breslow statistic: p=0·020)

Fig. 4—Life-table analysis of combined p21 and p53 status groups.p21" and p53" categories represent both negative and weaklypositive cases. p21"/p53+ (i.e. "ve/+ve in figure) n=16; +ve/"ven=22; "ve/"ve n=52. Only a single case was classified as +ve/+veand is excluded from the analysis. (Breslow statistic: p=0·008)

DISCUSSION

Using the technique of immunocytochemistry, thisstudy addresses the role of p21 protein expression indetermining both endocrine responsiveness and survivalin human breast cancer. p21-ICA status and otherbiological markers were determined on the primarytumour from patients who subsequently developed arecurrent lesion on which a response to endocrinetherapy could be determined. The patients in this studyhave been previously described18 and in these, ERnegativity, elevated Ki67, and EGFR positivity, but notp21ras or c-erbB2 protein, were associated with reducedtime to disease progression and overall survival. Asubsequent analysis of these data demonstrated asurvival advantage for p53-ICA negatives.

p21 staining was nuclear and heterogeneouslyexpressed in 82 per cent of cases with more than10 per cent of cells staining in 31 per cent of cases. Thesedata concur with Barbareschi et al.,16 whose study, usinga different antiserum and cut-offs, demonstrated anincreased disease-free interval for patients with highp21-expressing primary tumours, but found no relation-ships between p21 and ER, p53 or the proliferationmarker MIB1. A recent study,30 also using differentcut-off criteria, showed low p21 expression to correlatewith p53 overexpression, high tumour grade, and withrelapse-free and overall survival, but not with ER statusor tumour type.

In our study, p21 status appears inversely correlatedwith p53 status, but does not relate to ER or Ki67expression. p21 also did not correlate with response totherapy, or with TTP. A significant survival advantagedoes, however, appear to be afforded to those patientswhose tumours express elevated p21.

J. Pathol. 188: 126–132 (1999)


Table VI—Comparison of p21 and Ki67 immunocytochemical staining status using two positive groupsfor each

Ki67-negativeKi67-weakly

positiveKi67-strongly

positive Total

p21-negative 3 (3·3%) 4 (4·4%) 22 (24·2%) 29 (31·9%)p21-weakly positive 6 (6·6%) 7 (7·7%) 26 (28·6%) 39 (42·9%)p21-strongly positive 4 (4·4%) 6 (6·6%) 13 (14·3%) 23 (25·3%)Total 13 (14·3%) 17 (18·7%) 61 (67·0%) 91 (100%)


Fig. 5—Distribution of p21/p53 status groups by levels of Ki67expression. p21" and p53" categories incorporate both negative andweakly positive tumours. Horizontal bars represent mean Ki67 stain-ing levels for each group (Kruskal–Wallis one-way ANOVA p=0·169)

Fig. 6—Life-table analysis of the p21"/p53" cases only (n=52)which represents both negatives and weakly positives subclassifiedaccording to Ki67 status. Ki67+ +ves n=35; +ves n=8 and "vesn=9. (Breslow statistic: p=0·056)

Co-analysis of p21 and p53 data revealed the presenceof all four possible subgroups. The p21+/p53" pheno-type is considered to represent cells with normallyfunctioning wild-type p53 gene, the presence of p21protein being suggestive of normal p53-mediated tran-


scription via the p53-responsive element within the p21gene. Cells with such a phenotype might, as our dataindeed show, be expected to have a survival advantage,being capable, via p53 regulatory pathways, of efficientlymaintaining the integrity of their genome. In contrast,the p21"/p53+ phenotype is predicted to arise from amutated p53 gene transcribing a mutant p53 proteinwith an extended half-life and immunodetectability. Theapparent absence of p21 presumably results from aninability of the mutant p53 protein to initiate its tran-scription. Predictably, patients of this phenotype had theshortest overall survival and highest Ki67 levels.

Most patients in our study were p21"/p53" anddemonstrated intermediate levels of both cellular pro-liferation and overall survival. This group is likely tocomprise two subgroups: those with mutated p53 genesleading to an inability to translate p53 protein and thusto transcribe p21 by a p53-dependent pathway, andthose in which the p53 gene is wild type but whereinmost cells are quiescent, proliferation rates low, andwhere the requirement for p53 protein and cell-cycleinhibitory mechanisms would thus be minimal. Here p21levels would also be expected to be low. This assumptionof a mixed phenotype for this double-negative group issupported by our Ki67 data, which clearly differentiatea subgroup of tumours with low proliferation andgood survival characteristics similar to the p21+/p53"phenotype, from those with higher levels of proliferationand poorer survival characteristics more akin to thep21"/p53+ phenotype.

Only a single patient in our study was of the p21+/p53+ phenotype, reflecting a rare situation where highlevels of immunodetectable and normally functioningwild-type p53 protein are produced.

A number of studies have recently implicatedoestradiol and the oestrogen receptor as having regu-latory roles in p53 gene transcription.31,32 Hurd et al.,31

for example, have demonstrated oestradiol-inducedup-regulation of p53 protein in vitro. In our studypopulation, ER positivity was significantly associatedwith response, TTP, and overall survival. However, p21status failed to show any comparable correlation withER, endocrine response, or TTP, and thus p21-ICAassessment appears not to add any further independentdata for patient selection for endocrine therapy.

Recent studies, however, suggest that p21 may beimportant in mediating the actions of chemotherapy.For example, increased levels of p21 and apoptosis havebeen shown in wild-type p53-expressing cells in response

J. Pathol. 188: 126–132 (1999)


to DNA-damaging compounds such as adriamycin oretoposide,4,7,33 implying that p21 expression mightrender tumour cells more susceptible to such chemo-therapeutic agents. Although this concept is undoubt-edly oversimplistic, further clinical studies involvingchemotherapy-treated breast cancer patients shouldbetter resolve these issues and are assuredly underway.

ACKNOWLEDGEMENTS

We are grateful for the excellent technical assistanceof Mrs P. Finlay, Ms S. Kyme, and Mrs L. Farrow inthe preparation of the manuscript. We also continue tobe indebted to the Tenovus Organisation for theirgenerous financial support.

REFERENCES1. Lane DP, Crawford LV. T antigen is bound to a host protein in

SV40-transformed cells. Nature 1979; 278: 261–263.2. Lane DP. p53: guardian of the genome. Nature 1992; 358: 15–16.3. Kastan MB, Onyekwere O, Sidransky D, Vogelstein B, Craig RW.

Participation of p53 protein in the cellular response to DNA damage.Cancer Res 1991; 51: 6304–6311.

4. Clarke AR, Purdie CA, Harrison DJ, et al. Thymocyte apoptosis induced byp53-dependent and independent pathways. Nature 1993; 362: 849–852.

5. El-Deiry WS, Tokino T, Velculescu VE, et al. WAF1: a potential mediatorof p53 tumour suppression. Cell 1993; 75: 817–825.

6. Harper JW, Adami GR, Wei N, Keyomarsi K, Elledge SJ. Thep21 CDK-interacting protein Cip1 is a potent inhibitor of G1cyclin-independent kinases. Cell 1993; 75: 805–816.

7. El-Deiry WS, Harper JW, O’Connor PM, et al. WAF1/CIP1 is induced inp53 mediated G1 arrest and apoptosis. Cancer Res 1994; 54: 1169–1174.

8. Sheikh MS, Li XS, Chen JC, Shao ZM, Ordonez JV, Fontana JA.Mechanisms of regulation of WAF1/CIP1 gene expression in human breastcarcinoma: role of p53-dependent and independent signal transductionpathways. Oncogene 1994; 9: 3407–3415.

9. Leibermann DA, Hoffman B, Steinman RA. Molecular controls of growtharrest and apoptosis: p53-dependent and independent pathways. Oncogene1995; 11: 199–210.

10. Waga S, Hannon GJ, Beach D, Stillman B. The p21 inhibitor ofcyclin-dependent kinases controls DNA replication by interaction withPCNA. Nature 1994; 369: 574–578.

11. Barnes DM, Camplejohn RS. p53, apoptosis and breast cancer. J MammaryGland Biol Neoplasia 1996; 1: 163–175.

12. Juven T, Barak Y, Zauberman A, George DL, Oren M. Wild type p53 canmediate sequence-specific transactivation of an internal promoter within theMDM2 gene. Oncogene 1993; 8: 3411–3416.

13. Haupt Y, Maya R, Kazaz A, Oren M. Mdm2 promotes the rapiddegradation of p53. Nature 1997; 387: 296–299.

14. Fisher CJ, Gillett CE, Vojtesek B, Barnes DM, Millis RR. Problems withp53 immunohistochemical staining: the effect of fixation and variation in themethods of evaluation. Br J Cancer 1994; 69: 26–31.


15. Barbareschi M, Caffo O, Doglioni C, et al. p21WAF1 immunohistochemicalexpression in breast carcinoma: correlations with clinicopathologicaldata, oestrogen receptor status, MIB1 expression, p53 gene and proteinalterations and relapse-free survival. Br J Cancer 1996; 74: 208–215.

16. Barbareschi M, Doglioni C, Veronese S, et al. p21WAF1 and p53 immuno-histochemical expression in breast carcinoma may predict therapeuticresponse to adjuvant treatment. Eur J Cancer 1996; 32A: 2182–2183.

17. Nicholson RI, Gee JMW, Seery LT, et al. p53 expression in human breastcancer: relationship to tumour differentiation and endocrine response. In:Klijn JGM, ed. Prognostic and Predictive Value of p53. ESO ScientificUpdates Vol 1. Amsterdam: Elsevier Science, 1997; 1–20.

18. Archer SG, Eliopoulos A, Spandidos D, et al. Expression of rasp21, p53 andc-erbB2 in advanced breast cancer and response to first line hormonaltherapy. Br J Cancer 1995; 72: 1259–1266.

19. Andre S, Pereira H, Nogueira M, et al. p53 overexpression is a prognosticindicator in poorly differentiated, node negative and T1/T2 invasive ductalbreast cancer patients. Breast 1997; 6: 194–201.

20. Poller DN, Hutchings CE, Galea M, et al. p53 protein expression in humanbreast carcinoma: relationship to expression of epidermal growth factorreceptor, c-erbB2 protein overexpression and oestrogen receptor. Br JCancer 1992; 66: 583–588.

21. Ellis IO, Galea M, Broughton N, Locker A, Blamey RW, Elston CW.Prognostic pathological features in breast cancer. II. Histological type.Relationship with survival in a large study with long-term follow up.Histopathology 1992; 20: 479–489.

22. Elston CW, Ellis IO. Pathological prognostic factors in breast cancer. I. Thevalue of histological grade in breast cancer: experience from a study withlong term follow-up. Histopathology 1991; 19: 403–410.

23. Hayward JL, Carbonne PP, Heuson JC, Kumaoka S, Rubens R.Assessment of response to therapy in advanced breast cancer. Cancer 1977;39: 1289–1293.

24. British Breast Group. Assessment of response to treatment in advancedbreast cancer. Lancet 1974; ii: 175–178.

25. Midgley CA, Fisher S, Bartek J, Vojtesek B, Lane DL, Barnes DM.Analysis of p53 expression in human tumours: an antibody raised againsthuman p53 expressed in Escherichia coli. J Cell Sci 1992; 101: 183–189.

26. Walker KJ, Bouzubar N, Robertson J, et al. Immunocytochemicallocalization of oestrogen receptor in human breast tissue. Cancer Res 1988;48: 6517–6522.

27. Nicholson RI, Bouzubar N, Walker KJ, et al. Hormone sensitivity in breastcancer: influence of heterogeneity of oestrogen receptor expression and cellproliferation. Eur J Cancer 1991; 27: 908–913.

28. Locker AP, Birrell K, Bell JA, et al. Ki67 immunoreactivity in breastcarcinoma: relationships to prognostic variables and short term survival.Eur J Surg Oncol 1992; 18: 224–229.

29. Nicholson RI, McClelland RA, Finlay P, et al. Relationship betweenEGF-R, c-erbB2 protein expression and Ki67 immunostaining in breastcancer and hormone sensitivity. Eur J Cancer 1993; 29: 1018–1023.

30. Jiang M, Shao Z-M, Wu J, et al. p21/waf1/cip1 and mdm-2 expression inbreast carcinoma patients as related to prognosis. Int J Cancer 1997; 74:529–534.

31. Hurd C, Khattree N, Dinda S, Alban P, Moudgil VK. Regulation oftumour suppressor proteins, p53 and retinoblastoma, by estrogen andantiestrogens in breast cancer cells. Oncogene 1997; 15: 991–995.

32. Sheikh MS, Shao ZM, Hussain A, Fontana JA. The p53-binding proteinMDM2 gene is differentially expressed in human breast carcinoma. CancerRes 1993; 53: 3226–3228.

33. Lowe SW, Bodis S, McClatchey A, et al. p53 status and the efficacy ofcancer therapy in vivo. Science 1994; 266: 807–810.

J. Pathol. 188: 126–132 (1999)

p21waf1 expression and endocrine response in breast cancer

Documents