Br. J. Cancer (1974) 29, 232

FIBROSIS AS AN INDICATION OF TIME IN INFILTRATINGBREAST CANCER AND ITS IMPORTANCE IN PROGNOSIS

0. TH. ANASTASSIADES* AND D. M. PRYCEt*Front the Departmtent of I'athology, State General Hospital, Athens (Cholargos), Greece,

and thetDepartmient of Mlorbid Anatomy, University of London, St MUary's Hospital, London, England

Receive(d 5 April 1973. Accepted 28 November 1973

Summary.-The histological grading of tumours according to their intrinsicmalignancy is very important in the prognosis of breast cancer but within eachgrade the ultimate prognosis depends mainly on the age of the tumours.

We have shown that tumour fibrosis is an indication of this time factor, increasingwith the age of the tumour. Within each grade the metastatic ratio is higher andthe 5 year and 10 year survival less with the scirrhous than with the non-scirrhoustumours. The establishment of axillary metastases is closely connected with boththe degree of malignancy and the time available, the unfavourable effect upon survivalbeing greater in the scirrhous than in the non-scirrhous tumours.

Another consequence of the passage of time, as indicated by fibrosis, is thegradual diminution of lymphoid infiltration (LI) which is mostly present in youngtumours, especially those of high grades. The favourable effect of LI upon survivalis demonstrated in the non-scirrhous tumours of grade III, possibly because of itsgreat intensity, but this influence upon survival is lost as fibrosis increases and theintensity of the reaction diminishes.

REACTIVE fibrosis is a phrase oftenused in pathological reports. It is alsocommonly used by investigators of breastcancer. Gallager and Martin (1969) haveillustrated diagrammatically the centralfibrous reaction in infiltrating breastcarcinomata, referring to it as " reparativefibrosis ". Although the fibrosis in cancerappears to be truly reactive, this term isapt to convey the impression that fibrosisis a favourable feature. Evidence to thecontrary is, however, accumulating. Ina study of the connective tissue stroma ofbreast carcinomata Smolak, Kolodziejskaand Urban (1968) found that the 5 yearmortality was greatest with the poorlydifferentiated carcinomata with abundantand compact connective tissue stroma.Furthermore, Hamlin (1968) and Alderson,Hamlin and Staunton (1971) expressedsurprise at the high mortality score obtainedwith tumour fibrosis.

We have been interested in this matterfor several years. In a previous paper(Anastassiades and Pryce, 1966), in whichtumour size was used as an indication oftime, we realized that scirrhous tumoursrequired a separate scale because, beingshrunken, they corresponded with non-scirrhous tumours which were larger. Itseems probable therefore that on averagewithin each grade they would be olderand have a poorer prognosis. To put thismatter to the test, a new series of caseswas investigated. From the histologicalstudy of these cases we came to regardthe fibrous reaction as a continuous processtaking place during the evolution of br-eastcancer and we decided to evaluate itsinfluence on metastases and post-operativesurvival of the patients. We also decidedto assess the intensity of the lymphoidinfiltration (LI) of the tumours, thepresence of which has been proved

FIBROSIS AND ITS IMPORTANCE IN PROGNOSIS

beneficial by several authors (Moore andFoote, 1949; Black, Speer and Opler,1956; Richardson, 1956; Berg, 1959;McDivitt, Stewart and Berg, 1968; Cutleret al., 1969; Bloom, Richardson and Field,1970; Bloom, 1971) and to evaluate itspossible relationship to the degree ofmalignancy and fibrosis of the tumours andto the survival of the patients.

MATERIALS AND MIETHODS

The study was based on a series of 206consecutive cases of breast cancer operated onbetween the years 1945 and 1950 in St Mary'sHospital, London. Patients who had re-ceived preoperative radiotherapy were notincluded. The material for this study wastaken from the files of the Pathology Depart-ment of St Mary's Hospital and the survivaldata were obtained independently from thefollow-up files of the Department ofRadiology.

The original sections were examinedtogether with newr sections stained withhaematoxylin and eosin, Van Gieson stain andWeigert's elastin stain. In 8 cases noblocks could be found and in 3 others thefixation was too poor. As the purpose of thestudy wvas to make a straight comparisonbetween infiltrating scirrhous and non-scirrhous tumours, it was considered neces-sary to exclude 12 patients w%ith mucoidtumours and a similar number with muchintraductal growth but the minimum ofinfiltration Also discarded were 5 patientswith multiple tumours, a patient whosetumour had mutated and another in whom ithad spontaneously regressed. Three couldnot be used because there was no follow-upand 20 additional ones were discardedbecause the death of the patients was eitherunrelated to cancer or was due to an un-known cause.

The 141 cases used had a measurable massof infiltrating growth ranging from 0-5 to9-0 cm in greatest diameter. The number oflymph nodes in each case varied from 2 to 19.In 61 cases there were 2 to 5 lymph nodes; in42 the lymph nodes available were completelyreplaced by growth. The overall metastaticratio of the initially collected 206 cases was56% and of the 141 cases used 62%.

Grading of tumours was made accordingto Bloom and Richardson (1957). The

tumours were originally divided according tothe quantity of their fibre content into 4groups, but due to the paucity of numbers thefinal comparison was made betw%een thecombined two less, and the combined twNomore, fibrous groups. The combined lessfibrous group contained tumours with slightor little connective tissue stroma wAhich wrasfibroblastic and more or less evenly distri-buted throughout the wrhole tumour. Thecombined more fibrous group containedtumours with moderate or marked centralfibrosis with hvalinization, and correspondedto the well known scirrhous tumours.

The lymphoid infiltration (LI) at thecentre as well as at the periphery of thetumours was also estimated in each individualcase. In some tumours it w^as moderate ormarked whereas in others it was slight orabsent. The former two groups were regardedas positive and the latter two groups asnegative.

RESULTS

The collection comprised 30 tumours ingrade I, 60 tumours in grade II and 51tumours in grade III. The characteristicsof each individual tumour can be seen inthe scatter diagrams.

The metastatic ratio of the tumours inthe three grades was rather similar: 66%in grade I, 63% in grade II and 59% ingrade III. The overal 5 year and 10 yearsurvival, however, decreased with increas-ing grade, as can be seen in Table I.The decrease was even greater in themetastatic cases, as can be seen in TableII. These figures show that although themetastatic ratio is similar in all threegrades, prognosis is affected mainly bytumour grade and particularly whenmetastases are present, the greater en-hancement being in grade III.

The series contains 53 cases in thenon-scirrhous group and 88 in the scirr-hous group. It appears from the scatterdiagrams that the scirrhous tumotirs aremore frequent in grades I and II (70%o and710% respectively) and less frequent ingrade III (470%). They are also morefrequent among the metastatic cases (750o)compared with the non-metastatic cases(41 %)

233

0. TH. ANASTASSIADES AND D. M. PRYCE

SCI RRHOUS

0O 1 0 NON SCIRRHOUS

FIG. 1.-Thirty tumours of grade I are distributed according to their size and according to theirdegree of fibrosis on 2 cm scales. The tumours above each line are metastatic, and the tuinoursbelow the lines are non-metastatic.

0.

@0* 1 1 SCIRRHCUS

00 000V0i W 0'n

NON SCIRRHOUS

O&Gf OV i.)0O 0

FIG. 2.-Sixty tumours of grade II are distributed according to their size and according to theirdegree of fibrosis on 2 cm scales. The tumours above each line are metastatic and the tumoursbelow the lines are non-metastatic. Two tumours on the line have only minute metastases.

0 @*f4 0 0

I f * * * SCIRRHOUSM1 ~o or0~~~~

@ _ 0*:~ NON SCIRRHOUS

ocooVFIG. 3.-Fifty-one tumours of grade III are distributed according to their size and according to their

degree of fibrosis on 2 cm scales. The tumours above each line are metastatic and the tumoursbelow the lines are non-metastatic.

* SUCCUMBED WITHIN 5 YEARS5 YEAR SURVIVORS

o 10 YEARS SURVIVORS0 MODERATE LICr MARKED LI

234

FIBROSIS AND ITS IMPORTANCE IN PROGNOSIS

TABLE I.-Metastatic Ratio and Prognostic Outlook of the Whole Series, andScirrhous and Non-scirrhous Groups

Metastatic ratioGrade I 5 year survival

10 year survivalMetastatic ratio

Grade II 5 year survival10 year survivalMetastatic ratio

Grade III 5 year survival10 year survival

TABLE II. Prognostic Outlook of the Metastatic Cases of the Whole Series, andScirrhous and Non-scirrhous Groups

Grade I 5 year survival10 year survival

Grade II 5 year survival10 year survival

Grade III 5 year survival10 year survival

In each grade the prognostic outlookwas greatly worsened with increase offibre content. As can be seen in Table I,the metastatic ratio was greater, and the5 year and 10 year survival less, in thescirrhous than in the non-scirrhous group.It is therefore evident that the metastaticratio and the survival of the patients varyconsiderably among tumours of the samegrade according to their degree of fibrosis.

Among the metastatic tumours of thethree grades the prognostic outlook waseven worse in the scirrhous than in thenon-scirrhous tumours, as can be seen inTable II. The 5 year survival was less inthe scirrhous than in the non-scirrhoustumours in all three grades. The 10 yearsurvival data are not as convincing as faras grade II is concerned (possibly becauseof the small number of cases) as all the 6metastatic cases in the non-scirrhousgroup died within 10 years. However, ingrade I the 10 year survival was much lessin the scirrhous than in the non-scirrhousgroup and in grade III there were nosurvivors in the scirrhous group.

Lymphoid infiltration of the tumours(LI) was more frequent in the moremalignant grades. It was present in 10%

Overall60%35%,31%16%20%13%

Scirrhous53%290o28%18%0%

-Non-scirrhous100%660o50%0%46%30%

in grade I, 26% in grade II and 53%0 ingrade III. It seems therefore that themore malignant tumours evoke more LI.It was also more frequent in the non-scirrhous group in all three grades, as canbe seen in Table III. Although morecommon in the non-scirrhous tumours, itseems important to note its existence in aconsiderable proportion of the scirrhoustumours in the two more malignantgrades, especially in grade III. But as canbe seen in the scatter diagram for thescirrhous tumours of this grade, LI waspresent predominantly in moderate degreecompared with the non-scirrhous tumoursof the same grade, where it was presentpredominantly in marked degree.

TABLE III. Incidence of LI in the WholeSeries, and in Scirrhous and Non-scirrhous Groups

Overall Scirrhous Non-scirrhouisGrade I 10% 0% 33%Grade II 26% 19% 47%Grade III 53% 42% 63%

The influence of LI on metastases andsurvival of the patients is particularlyevident in the non-scirrhous tumours of

Overall66%66%46%63%43%30%59%39%31%

Scirrhous80%57%33%74%37%28%70%20%16%

Non-scirrhous33%89%77%35%58%35%48%55%44%

235

0. TH. ANASTASSIADES AND D. M. PRYCE

grade III, as can be seen in the scatterdiagram for this grade. Eleven out of 17(64%) LI positive cases were non metas-tatic. Also, 12 out of 17 (70%0) LI positivecases survived a years and 9 cases (53%0)survived 10 years. The influence of LI ingrade I cannot be considered because of thesmall number of LI positive cases. Ingrade II the results are irregular and noinfluence of LI can be demonstrated.

DISCUSSION

The close relationship between gradeand prognosis has been repeatedly reportedby many authors (Black and Speer, 1957;Bloom and Richardson, 1957; Bloom,1962, 1965, 1971; Cutler et al., 1966; Wolf,1966; Hamlin, 1968; Tough et al. 1969,Alderson et al., 1971). Accordingly, thedata from this series of cases show that the5 year and 10 year survival decreases withincreasing grade.

Although the histological grading oftumours (Bloom and Richardson, 1957;Black and Speer, 1957) is an artificial andsubjective procedure, it appears to be ofgreat clinical value because it roughlyreflects the rate of growth of the neoplasticcells and consequently the rapidity withwhich the disease progresses.

Nevertheless, the tumours of the threegrades in our series of cases exhibitsimilar metastatic ratios (grade I 66%,grade II 63%, grade III 59%0). Compar-able results have also been reported byBell, Friedell and Goldenberg (1969).Kreyberg and Christiansen (1953) have alsopointed out that the metastatic ratio ofgrade I tumours is as high as that of themore malignant grades. The fact that, in agiven series of cases, slow growing andrapidly growing neoplasms exhibit asimilar metastatic ratio, shows that theestablishment of the regional lymph nodemetastases is a result of the competitiveinterplay of more than one factor. Amongthese the chronological age of the tumoursand the reactivity of the host have beenconsidered of potentially great importancetogether with the rate of growth of theneoplastic cells.

The rate of growth of the neoplasticcells, differing considerably among thevarious tumours (Slack et al., 1969;Kusama et al., 1972), can be roughlyevaluated histologically by the variousgrading systems, but the chronological ageof any given tumour cannot be assessedby any of the known procedures. Thedelay of the patients in seeking treatmenthas been used as indirect evidence byBloom (1965), who has demonstrated thatit exerts an influence on survival when thegrade of the tumours is taken into accountand the delay refers to tumours of similarrate of growth. It is obvious that whenindolent, slow growing tumours are dis-covered they should be smaller and presentfor longer time than rapidly growingaggressive tumours.

In our series of cases, grade I tumoursare as a whole of smaller size than those ofgrade III tumours, 76% of grade I and53% of grade III being 3 cm or less at theirgreatest diameter. Grade II is the mostartificial group of tumours, sharing charac-teristics of both borderline grades, theproportion of small tumours in thisgrade being 65%.

The preponderance of small tumours ingrade I (76%) is indicative of their slowrate of growth compared with the tumoursof the other grades. But the metastaticratio of these small tumours, in spite oftheir low malignancy, is very high (60%),probably because of their long existence.It is even higher than the metastatic ratioof the most aggressive grade III tumours ofthe same size (510%). It appears, there-fore, that the age of the tumours interfereswith aggressiveness, with the result that thethree grades exhibit a similar metastaticratio. But in spite of the fact that gradeI tumours are older their overall r5 yearand 10 year survival is greater than in theother grades (5 year survival 66% com-pared with 43%0 and 39%0 in grades II andIII, and 10 year survival 46% comparedwith 30 and 31% in grades II and IIIrespectively). This emphasizes the enor-mous prognostic importance of the degreeof malignancy of the tumours, which

236

FIBROSIS AND ITS IMPORTANCE IN PROGNOSIS

constitutes a dominant feature in breastcancer, predetermining the rapidity withwhich the evolution of the disease will takeplace.The time factoor

From the study of our series of cases,we came to the conclusion that someevidence of tumour age could be suggestedby histological changes found in thetumours themselves. The various appear-ances in breast cancer are still described intextbooks of pathology as immutable andno consideration appears to have beengiven to the possibility that histologicalchanges occurring with the passage of timecould be evaluated and used for prognosis.Such changes are taking place in alltumours and are more or less constant.In some infiltrating breast cancers theconnective tissue stroma is sparse andfibroblastic. in others more abundant andin still others there is an enormousincrease of the connective tissue stromawith distinct phenomena of maturationand scarring. The increase and sclerosisof the connective tissue stroma whichhistologically are undoubtedly ageingprocesses are found as one proceeds fromthe periphery to the centre of the tumour,going from the more recent to the olderparts of it. The sclerotic centre varies inextent in the various tumours. It iswell developed in tumours of low degree ofmalignancy probably because of their longduration. Although less extensive, it isalso found in tumours of intermediate andhigh degrees of malignancy, reflecting theirlonger duration compared with tun4ours ofthe same degree of malignancy with scantyand fibroblastic stroma.

In our series of cases we found thatthere is a significant difference in themetastatic ratio and the survival ofthe patients between the scirrhous andthe non-scirrhous tumours in all threegrades (Table I).

Although the separation of tumoursaccording to their degree of fibrosis in twogroups is an artificial division as thedevelopment of fibrosis is a continuous

process, the greater metastatic ratio of thetumours and the less survival ofthe patientsof the scirrhous groups compared with thenon-scirrhous groups in all three grades,reveal the prognostic significance of thetime factor among tumours of similar rateof growth. It is also clear that tumours ofthe same grade cannot be taken as ahomogeneous group because of the widespectrum of ages among the population ofeach grade. The preponderance of thescirrhous tumours in grade I (70%0) andthe high (80%)0 indeed the highest,metastatic ratio of these tumours (al-though they are slow growing) indicatethat they are the oldest group of tumoursand that they have the greatest timescale. But the prognostic significance ofthe time factor, as evidenced by fibrosis, isgreatest for the most malignant gradeIII tumours although they have a shortertime scale. The scirrhous tumours of thisgrade have the least 5 year and 10 yearsurvival (20%0 and 16oo respectively).

These findings show that the degree ofmalignancy in breast cancer constituteswhat Bloom (1971) precisely characterizedas the inherited " tempo " of the diseaseand that the time available, as indicatedby the degree of fibrosis, permits theinherited aggressiveness to kill the patientsin an increasing rate with increasingmalignancy.

The influence of the age, together withthe malignancy of the tumours, uponsurvival is most in evidence in cases inwhich metastases have already developedand both factors are decisively active(Table II). In these cases the survival ofthe patients is even worse compared withthe whole series. The metastatic gradeIII tumours with much fibrosis are, aswould be expected, the tumours with theworst prognosis. All the patients of thisgroup died within 5 years.The host reactivity

There is already a large amount ofevidence of host resistance of an im-munological nature in breast cancer,including reaction in the tumours them-

-37

238 0. TH. ANASTASSIADES AND D. M. PRYCE

selves and in the regional lymph nodes.The presence of LI in the primary tumourshas been repeatedly reported as accom-panied by better prognosis. The relation-ship of this type of host reactivity to thedegree of malignancy and fibrosis of thetumours requires further consideration.

The data from our series of cases showthat LI is very common in breast cancer,particularly in the more malignant gradeswhich may be more antigenic, as hasalready been pointed out by Bloom et al.(1970). However, it appears that thisreactivity is more characteristic of tumourswhich are relatively young and have lessfibrosis in all three grades (Table III).There is therefore a decrease in theincidence of LI as fibrosis progresses,which is accompanied by its gradualdisappearance from the centre of thetumours. Otherwise the disappearance oflymphocytes and plasma cells is a normalphenomenon when maturation and sclero-sis of connective tissue is taking place.

The influence of LI on survival is mostin evidence in the non-scirrhous tumoursof Grade III, in which it is found in itshigher incidence and intensity.

With the highly malignant non-scirr-hous tumours of grade III the results arein agreement with the previous authorswho have shown that medullary tumourswith LI have a good prognosis in spite oftheir high intrinsic malignancy.

As can be seen in the scatter diagramfor the non-scirrhous tumours of grade III,12 out of 17 LI positive cases (70%/)survived 5 years and 9 cases (52%/)survived 10 years. From the 10 LInegative cases only 3 (300%) survived, thesurvival exceeding 10 years. This differ-ence in prognosis between the LI positiveand LI negative non-scirrhous tumours ofgrade III must be due to host reactivity.But with increasing fibrosis of the tumourshost reactivity, as expressed by LI,lessens and finally ceases and its influenceon prognosis is completely lost in thescirrhous grade III tumours, although LIstill persists in moderate degree in some ofthese tumours. As can be seen in the

scatter diagram for the scirrhous tumoursof grade III, only one of the 10 LI positivecases survived.

This probably means that the passageof time gradually neutralizes the exertedbeneficial effect of host resistance of thistype. If the non-scirrhous LI positivetumours of grade III were not excised atthis comparatively early stage of theirevolution they would become examples ofthe scirrhous group of the same grade andhave high mortality. Fortunately, how-ever, these tumours in spite of theirtendency to be large are young, and evenyounger than the non-scirrhous tumoursof grade II, just as these are younger thanthe non-scirrhous tumours of grade I.Furthermore, the rapid increase in size ofthese tumours makes the clinical diagnosispossible at this earlier stage of theirdevelopment with the reactivity of thehost to the tumour, as expressed by LI,still being active.

We are indebted to the PathologyDepartment of St Mary's Hospital forproviding the material for this study andfor the technical assistance and to Mr P.Lagogiannis for his help in the preparationof the diagrams.

REFERENCESALDERSON, M. R., HAMLIN, I. M. E. & STAUNTON,

M. D. (1971) The Relative Significance of Prog-nostic Factors in Breast Carcinomas. Br. J.Caancer, 25, 646.

ANASTASSIADAS, 0. TH. & PRY(E, D. M. (1966)Immunological Significance of the MorphologicalChanges in Lymph Nodes Draining Breast Cancer.Br. J. Cancer, 20, 239.

BELL, J. R., FRIEDELL, G. H. & GOLDENBERG, I. S.(1969) Prognostic Significance of PathologicalFindings in Human Breast Carcinoma. SurgeryGynec. Obstet., 129, 258.

BERG, J. W. (1959) Inflammation and Prognosis inBreast Cancer. A Search for Host Resistance.Cancer, N. Y., 12, 714.

BLACK, M. Al. (1970) Human Breast Carcinoma.Part II. Research Potential. N.Y. St. J. Med.,15, 962.

BLACK, M. M., SPEER, F. D. & OPLER, S. R. (1956)Structural Representation of Tumor-Host Rela-tionship in Mammary Carcinoma. Am. J. clin.Path., 26, 250.

BLACK, M. M. & SPEER, F. D. (1957) NuclearStructure in Cancer Tissues. Surgery Gynec.Obstet., 105, 97.

FIBROSIS AND ITS IMPORTANCE IN PROGNOSIS 239

BLOOM, H. J. G. & RICHARDSON, W. W. (1957)Histological Grading and Prognosis in BreastCancer. Br. J. Cancer, 11, 359.

BLOOM, H. J. G. (1962) The Roles of HistologicalGrading in the Study of Breast Cancer. Sym-posium on the Prognosis of the Malignant Tumoursof the Breast. Ed. P. Denoix and C. Rouquette.Basel: Karger. p. 51.

BLOOM, H. J. G. (1965) The Influence of Delay onthe Natural History of Breast Cancer. Br. J.Cancer, 19, 228.

BLOOM, H. J. G., RICHARDSON, W. W. & FIELD, J. R.(1970) Host Resistance and Survival in Carcinomaof the Breast. A Study of 104 Cases of MedullaryCarcinomas in a Series of 1411 Cases of BreastCancer Followed for 20 years. Br. med. J., ii,181.

BLOOM, H. J. G. (1971) Impact of Tumor Grade andHost Resistance. Cancer, N. Y., 28, 1580.

CUTLER, S. J., BLACK, M. M., FRIEDELL, G. H.,VIDORE, R. A. & GOLDENBERG, I. S. (1966)Prognostic Factors in Cancer of the FemaleBreast. II Reproducibility of HistologicalClassification. Cancer., N. Y., 19, 75.

CITTLER, S. J., BLACK, M. M., MORK, T., HARVEI, S.& FREEDMAN, C. (1969) Further Observations onPrognostic Factors in Cancer of the FemaleBreast, Cancer, N. Y., 24, 653.

GALLAGER, H. S. & MARTIN, J. E. (1969) EarlyPhases in the Development of Breast Cancer.Cancer, N. Y., 24, 1170.

HAMLIN, I. M. E. (1968) Possible Host Resistance inCarcinoma of the Breast. A Histological Study.Br. J. Cancer, 22, 383.

KREYBERG, L. & CHRISTIANSEN, T. (1953) The Prog-nostic Significance of Small Size in Breast Cancer.Br. J. Cancer, 7, 37.

KuSAMA, S., SPRATT, J. S., DONEGAN, W. L.,WATSON, F. R. & CUNNINGHAM, C. (1972) TheGross Rates of Growth of Human MammaryCarcinoma. Cancer, N. Y., 30, 594.

McDIVITT, R., STEWART, F. & BERG, J. W. (1968)Tumors of the Breast. Atlas ofTumour Pathology.Armed Forces Institute of Pathology.

MOORE, 0. S. & FOOTE, F. W. (1949) The RelativeFavourable Prognosis of Medullary Carcinoma ofthe Breast. Cancer, N. Y., 2, 635.

RICHARDSON, W. W. (1956) Medullary Carcinoma ofthe Breast. Br. J. Cancer, 10, 415.

SLACK, N. H., BLUMENSON, L. E. & BROSS, I. D. J.(1969) Therapeutic Implication from a Mathe-matical Model Characterizing the Couirse ofBreast Cancer. Cancer, N. Y., 24, 960.

SMOLAK, K., KOLODZIEJSKA, H. & URBAN, A. (1968)Investigation on the Relationship Between theClinical Course of Mammary Carcinoma and theMorphological Features of the Stroma. Pol.med. J., 7, 162. (Translated from NowotworyXVII, No. 2/1967.)

TOUGH, I. C. K., CARTER, D. C., FRAZER, J. &BRUCE, J. (1969) Histological Grading of BreastCancer. Br. J. Cancer, 23, 294.

WOLFF, B. (1966) Histological Grading in Carcinomaof the Breast. Br. J. Cancer, 20, 36.