The Impact of Pathology on the Biologic, Diagnostic, Prognostic, and Therapeutic Considerations in Breast Cancer

Symposium on Breast Cancer

The Impact of Pathology on the Biologic, Diagnostic, Prognostic, and Therapeutic Considerations in Breast Cancer Edwin R. Fisher, M.D.*

ACKNOWLEDGMENT

Author's Note: During my formative years in pathology, I had the good fortune to be associated with Dr. George Crile. He provoked in me not only a desire to learn, but also, more importantly, the courage to think and to challenge--characteristics that I hope are reflected in the following article.

Conceptual aspects of most diseases in medicine-such as breast cancer-have been notoriously rigid. Historically, practitioners have been resistant to change. Thus the dramatic revisions that have occurred in the last decade-a period approximately encompassing one-tenth of the "modern" history of this disease-in views relating to the biology and treatment of this disease are no less than astonishing. It is true that the seeds of such change had been sown prior to this renaissance, but those efforts seemed to be isolated events; furthermore, in most instances they were greeted with skepticism because they either failed to embody scientific principles in the formulation of their conclusions or were based on experimental studies of which practitioners were unaware or were reticent to explore in the clinical setting. These objections were dispelled in large measure by the use of prospective randomized clinical trials, the scientific method of clinical problem-solving. The controlled nature of these clinical trials has also allowed for more meaningful pathologic observations. It became readily appreciated that an understanding of the pathologic aspects of breast cancer is essential to the planning of many clinical trials. The clinical findings obtained from such studies unquestionably served as a stimulus for other relevant pathologic investigations. They did not, however, always resolve all the long-standing *Director, Institute of Pathology, Shadyside Hospital, Pittsburgh, and Professor of Pathology, University of Pittsburgh School of Medicine, Pittsburgh, Pennsylvania This work was supported by USPHS Grant R-01-CA-12027 and Contract N01-CB-23876 and by American Cancer Society Grant RC-13.

Surgical Clinics of North America-Yo!. 64, No.6, December 1984

1073

1074

EDWIN

R.

FISHER

pathologic problems associated with the diagnosis of breast diseases. Too, findings in some instances were often hailed as inviolable evidence against the wisdom of performing some clinical studies, or at the least may tend to obfuscate such attempts. This article addresses some of the pathologic problems that have a bearing on the biologic, diagnostic, prognostic, and therapeutic aspects of breast cancer. THE CONCEPT OF EARLY AND MINIMAL BREAST CANCER It is unfortunate that many continue to regard the biology of breast cancer as a relatively stereotyped, time-related progression of events. This view portrays a cancer that grows with time and subsequently spreads to the regional lymph nodes, and then, with time, spreads systemically; therefore, interruption of the cycle before the cancer spreads to the regional lymph nodes or prior to systemic dissemination from these structures should effect a cure. Such a belief ratifies to many the value of so-called early detection and/or treatment. Although it cannot be denied that primary tumors take time to grow, relating regional and subsequently systemic spread to time is, according to more moderate tenets and experiences, not only misleading but inaccurate. That breast cancer represents a heterogeneous group of neoplasms is now obvious, and recent evidence suggests biologic heterogeneity of cells comprising individual breast cancers. Despite this lack of uniformity, there are certain principles that apply to the growth of cancers, including those of the breast, which further provoke skepticism concerning clinical designations of "early" or "late." It is well appreciated that most breast cancers cannot be palpated until they are at least 1 em in size, or l billion cells. Kinetic studies indicate that such a size requires 30 population doublings. When it is recognized that a doubling time might encompass 30 to 200 or more days, it becomes apparent that a tumor that is regarded as clinically early is in truth biologically late, requiring only 10 to 20 more doublings before causing death of the host. A small 0.5-cm breast cancer detected by mammography-although regarded as early clinically-has already traversed through 27 doublings, and is biologically a late tumor. The most important consideration, of course, is when a tumor, if it does contain metastasizing phenotypes, exhibits such a phenomenon. Most evidence suggests that when metastases occur, they do so within the first 10 to 20 doubling times, or at a stage undetectable by prevailing methodologies. It has been estimated that 50 per cent of women with a breast cancer measuring l em already have systemic disease. It is difficult to rationalize the experiences noted in some women with untreated breast cancer with the concept of early breast cancer or the pejorative effects of delay in diagnosis and treatment. Thirty-five per cent of women with untreated breast cancer have been noted to survive for 5 years. Sixty-eight per cent who presented with localized disease and were untreated survived at least this long. 46 Certainly, the results obtained in Protocol4 of the National Surgical Adjuvant Breast Project (NSABP) refutes the significance of delay in diagnosis and surgical treatment on survival. No

THE IMPACT OF PATHOLOGY ON BREAST CANCER

1075

difference in disease-free survival has been observed in ·this cohort of patients when these latter patients were grouped according to duration of symptoms, that is, 0 to 3, 3 to 6, 6 to 9, and longer than 9 months. 23 Further, one subset of patients with clinically negative nodes was subjected to just total mastectomy without intervention of the ipsilateral axilla. 12 Probably 40 per cent of the patients in this subset had positive nodes, an estimate based on the recognized error in clinical as opposed to pathologic staging of patients with negative nodes. 19 Yet, surprisingly, only 16 per cent had evidenced axillary recurrence within 10 years after mastectomy, and the treatment failure rate in the tenth postoperative year for this subset is statistically similar to that of other subsets who were subjected to either radical mastectomy or irradiation following total mastectomy. In keeping with this observation is the recognition of comparable survival rates in patients with occult micronodal metastases and those lacking such a phenomenon. 24 • 26 These considerations indicate one of the most salient shortcomings of the conventional concept of tumor biology-the failure to recognize the significance of host response and tumor characteristics in its equation. Our own experimental studies clearly demonstrated that the so-called barrier function of lymph nodes, a pivotal feature in the conventional view, is incorrect. Tumor cells may traverse lymph nodes either through the efferent lymph channel or lymphaticovenous communications, which, as well as passage through interstitial spaces, occurs with alarming alacrity. 9-u In our view, the removal of lymph nodes is performed to properly stage the neoplastic disease rather than to remove tumor per se and thwart subsequent metastases. This has led to an alternate hypothesis, 8 which emphasizes that the positive regional lymph node is not an instigator of tumor cell dissemination but an indicator of a host-tumor interrelationship that permits the development of metastases. Regional lymph nodes are of biologic rather than mechanical importance. The blood vascular system is an important pathway of tumor dissemination. It is, however, closely interrelated to the lymphatic system. The complex (and to date, often unclear) host-tumor relationship, which probably affects every facet of the disease, must be operative. In a large proportion of instances, operable breast cancer is already a systemic disease, so it is unlikely that variations in locoregional treatment will affect survival rates-a caveat that is historically correct. Certainly it is good medical practice to treat disease promptly. Yet, it must not be done under the guise that this promptness in the clinical setting invariably represents or plays a deciding role in the outcome. Theoretically, elimination of tumor burden is highly desirable if combative host factors or chemotherapeutic agents are to achieve their maximum efficiency. It remains to be demonstrated whether this latter factor will exert a beneficial effect, even in the adjuvant setting for all patients or just certain subsets in accord with the heterogeneity of breast cancers. Finally, until survival rates can be demonstrated to be substantially improved in prospective, randomized clinical trials, with the removal of tumors regarded clinically as early, it would appear that recognition of small tumors may have its greatest virtue in allowing for more cosmetic operative intervention rather than for biologic reasons. Results of the HIP mammographic screen-

1076

EDWIN

R.

FISHER

ing study is relevant in this regard. A survival rate of better than 30 per cent was noted only for women older than 50 years. 53 The possibility that this finding might have been influenced by length or lead time biases59 has not been resolved. These considerations also appear relevant in regard to the use of the term "minimal cancer,"38 an expression that connotes curability as well as earliness. By definition, minimal cancer includes lobular carcinoma in situ, intraductal cancers, and invasive cancers measuring no larger than 0.5 em. There is no biologic reason why in situ cancers should not be curable, but I believe that relatively few invasive cancers ::5 0.5 em will be encountered with available detection methods. Further, there are other cancers, which exhibit a favorable clinical course, that do not qualify as minimal cancer. Mucinous, tubular, and papillary forms of breast cancer are frequently larger than 0.5 em, yet they exhibit a very favorable prognosis. Because of these inconsistencies, it is preferable to refrain from using the general term minimal and to regard precise designations or descriptions of a particular lesion as more meaningful. The above indicates that the identifying features of what constitutes an early cancer have not been well defined, at least in the present state of the art. Conversely, there are certain mammary changes that have long been suspected as representing precursor lesions. Although these, too, defy a precise connotation of time, it is more certain that they occur both morphologically and biologically earlier than invasive cancer. One such lesion is the proliferative form of fibrocystic disease, or that mammary alteration characterized by ductal hyperplasia, vis-a-vis papillomatosis, epitheliosis, and multiple papillomas. Our own electron microscopic studies 15 of such lesions (including those designated as atypical hyperplasia because the delineation between hyperplasia and intraductal cancer might be regarded as equivocal) disclosed the banal form of hyperplasia to be composed of normal ductal epithelial cells, whereas the atypical form contained many, but not all cells, with features indistinguishable from those found in overt invasive carcinomas. More objective evidence concerning the relationship between banal hyperplasia and carcinoma was obtained from experimental studies performed with 3-methylcholanthrene (MCA) in syngeneic rats. 21 Karyotypic abnormalities including marker forms were noted in ductal cells from all phases of this process, with the least number occurring in cells from banal hyperplasia and the greatest in the cancers. Varying incidences of aneuploidy were also observed with cultured cells from human hyperplastic lesions and cancer. 25 Studies by Guillino and associates, 40 • 43 using a different model system, complement these findings. The findings from both studies represent compelling evidence that at least the proliferative form of fibrocystic disease represents not only a precursor of mammary carcinoma, but perhaps one of its earliest morphologic expressions. The failure to find karyotypic aberrations in all samples of proliferative fibrocystic disease, or as in the case of Gullino, 43 angiogenesis following xenografting to the anterior chamber of the rabbit's eye, does not militate against such an interpretation. Lesions indistinguishable from those of early hyperplasia in the MCA model may be induced by the administration of estradiol or progesterone.

THE IMPACT OF PATHOLOGY ON BREAST CANCER

1077

Yet hyperplastic lesions in this circumstance are not noted to develop- into lesions of advanced hyperplasia or cancer and lack the aneuploidy found in those following MCA. 20 Thus, not all hyperplastic lesions are expected to evolve into cancer. The arrest of such evolution appears to occur prior to the development of the advanced or atypical stage (lesions that possess only subtle light and electron microscopic, as well as karyotypic, differences, from those designated as cancer). These experiments indicate that the evolutionary process is complicated and certainly nonobligate. Unfortunately, there are no simple methods available that will allow identification of those hyperplastic lesions that will represent the truly high-risk forms of hyperplasia, and thus estimates of the degree of risk attendant with such lesions may be inaccurate. The distinction between what may be generally designated as intraductal hyperplasia and intraductal carcinoma of the breast remains difficult. There are some criteria, but the decision in some instances often still appears to many to be in large part subjective. In our practice, we utilize a mixed topographic and cytologic algorithm to differentiate between intraductal hyperplasia and carcinoma. 16• 17• 36 Intraductal lesions with overt cytologic atypia, and/or comedo necrosis, and/or signet ring change are regarded as carcinomatous (with the caveat that, on very rare occasions, necrosis may be encountered in a solitary small duct comprised of cytologically well-differentiated cells in a setting of ductal and lobular change that is obviously banal). Most difficulty occurs in intermediate and larger ducts whose lumens are composed of a surfeit of cells that are only slightly or not at all atypical. Distorted, angulated lumens occur. This is often referred to as a "busy" or "worrisome" appearance. Such lesions are regarded as hyperplastic, as the luminated structures exhibit a distinctly slip-like configuration. Yet those intraductal proliferations in which the lumens are represented by crisply formed "holes" (adenocystic configuration) rather than slits are considered as intraductal carcinomas, regardless of the degree of cytologic atypia of its component cells. Such a pattern may be encountered solely at the periphery of an otherwise empty, dilated duct, or among excessive intraductal cellular elements. The term atypical intraductal hyperplasia is reserved for those instances in which the observer has difficulty in concluding whether the lumens are "slits" or "holes." The same criteria are employed for intraductal proliferations of apocrine and papillary types. Papillary lesions are especially difficult, because, as noted previously, the presence of a so-called stalk, contrary to the popularly held view, does not exclude a diagnosis of papillary cancer. 28 Admittedly, this approach is relatively simple and unsophisticated, yet it does appear to represent a crystallization of the prolix that has been offered to delineate these lesions (Figs. 1 through 4). Further, this formula has been found to be highly reproducible. Recently, attention has been directed to another mammary lesion that may be highly significant in the histogenesis of breast cancer. It is characterized as a distinct, radial, stellate, or puckered scar in breast tissue with varying degrees of duct ectasia, mazoplasia, papillomatosis, and occasionally true papillomas at its periphery (Fig. 5). The central portion of the lesion is composed of dense hyaline, with an abundance of elastic

1078

EDWIN

R.

FISHER

Figure 1. Intraductal hyperplasia. The luminated spaces are slit-like.

Figure 2. Intraductal carcinoma. In this form, the lumens are crisp, punched-out holes.

• THE IMPACT OF PATHOLOGY ON BREAST CANCER

1079

Figure 3. Atypical intraductal hyperplasia. The lumens are neither clearly slit-like nor definite holes.

Figure 4. Intraductal carcinoma signet ring type. This is a frequently overlooked form of intraductal carcinoma.

1080

EDWIN

R.

FISHER

Figure 5. Non-encapsulated sclerosing lesion (NESL) of breast. Arrow points to central core. Radiating spokes are evident as is peripheral duct ectasia.

fibers as well as tubules. The latter often exhibit reduplication of lumens and frequently appear to develp by budding from larger ducts exhibiting papillomatosis. The cells comprising the ducts within the central core are well differentiated. Mitotic figures are not evident or are exceedingly rare. Myoepithelial cells, as well as histologic basement membranes, are variably observed. This constellation of findings has been designated by a number of appellations but we have utilized the descriptive term "nonencapsulated sclerosing" lesion (NESL). 27 The significance of this lesion is tWofold: (1) such lesions are often worrisome to pathologists who find them difficult to distinguish from tubular carcinoma; and (2) there is evidence that suggests that the NESL may represent a precursor of tubular and other carcinomas of the breastY· 45 We have failed to discern any objective histologic or histochemical difference between the tubules comprising the NESL and tubular cancers. Ultrastructural examination of an example of NESL also revealed a commonality of cellular features. Large ducts within some of the lesions may exhibit unequivocal intraductal carcinoma. Further, we, 33 as well as others, 45 have observed what appears to be bona fide transitions between the NESL and small incipient(?) tubular and other cancers (Fig. 6). Some of these have been noted to exhibit axillary nodal metastases. The decision whether a lesion represents tubular cancer or simple NESL is often made by recognizing true infiltration of the tubular elements into surrounding adipose tissue. Such invasion is characterized by "naked" tubules within the fat, as opposed to pseudoinfiltration, which is character-

THE IMPACT OF PATHOLOGY ON BREAST CANCER

1081

Figure 6. Scar cancer, type l. Remnant of the central core is evident (arrow), as are radiating spokes similar to that of banal NESL in Figure 5. Carcinomatous elements are clearly evident at the periphery.

ized by hyaline or fibrous sheaths about the tubular elements within the fat. A review of pathologic materials from 1569 women enrolled in NSABP Protocol 4 revealed that the invasive cancers in 38 per cent of the women exhibited features of scar13 (the term scar is used to connote a somewhat distinct stromal reaction and is in no manner related to trauma or previous breast biopsy; actually, five types of scar could be identified morphologically). However, in 18 per cent of patients with scar or 7 per cent of patients of the entire series, the latter exhibited features with that observed in the banal NESL described above (Fig. 5). Such cancers were designated as type 1 scar cancers. There was no evidence from that study to indicate that type 1 scar cancers, which were usually well differentiated, progressed into more anaplastic forms. Interesting from a pathogenetic standpoint was the recognition of vascular change in scar cancers and their similarity to socalled scar cancers of the lung. MULTICENTRICITY AND BILATERALITY

That multicentric foci of cancer, that is, the presence of cancer in quadrants remote from the dominant mass, do occur is no longer a subject for debate; indeed, estimates indicate their presence in 14 to 40 per cent

1082

EDWIN

R.

FISHER

of invasive breast cancers. The association of this phenomenon with lobular carcinoma in situ (LCIS) has been stated by some to be as high as 70 per cent. 1 Yet, the most compelling consideration concerns the biologic significance of such multicentric foci, which are in the vast majority of instances represented by noninvasive LCIS and intraductal cancers. Studies indicating the development of cancer in one-third of patients with LCIS 51 and even higher numbers of patients with intraductal cancer treated by local excision has represented the principal reason for the fallibility of such primary surgical treatment of breast cancer. Yet, it should be recognized that the number of cases on which such a conclusion is based is small; that, in instances of LCIS, the occurrence of contralateral cancer is equal to that of cancer in the ipsilateral member; that the studies are retrospective without concurrent controls; and that, most important, it is unclear whether the developing cancer occurred at the previous biopsy site (which would represent incomplete excision or multifocal rather than multicentric cancer). Indirect evidence indicates that such cancers may not be as clinically important as proposed. First, the incidence of second cancers arising after conservative therapy for LCIS or intraductal cancer falls far short of the purported 70 per cent. This information coincides with studies indicating that such cancers in situ are 19 times more frequent than clinical cancers in the breasts of women of older age groups studied at necropsy. 44 Despite the relatively high frequency of multicentric foci in patients with invasive cancer, the occurrence of two dominant mass cancers in the same breast is exceedingly rare, being estimated to be 0.1 per cent, 19 much less than the incidence of contralateral cancer. There are very few contemporary surgeons who would suggest prophylactic contralateral mastectomy in the treatment of breast cancer. It is unfortunate that, in the one prospective study in which quadrantectomy was performed for invasive cancer, all patients were treated with postoperative x-ray. 57 This negated the opportunity for studying the biologic significance of multicentric foci. However, it is of interest that recurrent cancer was found in only 1 per cent of the cases treated by quadrantectomy, an incidence that was even less than local recurrence following radical mastectomy. No data were presented in that study to indicate where the recurrence appeared. Ultimate resolution concerning the significance of multicentric cancers will be derived from the NSABP Protocol 6, which represents a threearmed prospective clinical trial with patients receiving either segmental mastectomy (lumpectomy) with and without postoperative irradiation or modified radical mastectomy. The protocol has been recently closed to patient entry, so that information available must be regarded as very preliminary. Results thus far suggest a local tumor recurrence rate of approximately 5 to 10 per cent. All such recurrences to date have appeared within the same region of breast as the primary excision, indicating either incomplete excision or true multifocality (the term multifocal indicates tumor in the same quadrant or vicinity of the dominant mass, as opposed to multicentricity as defined above). The majority oflocal breast recurrences have occurred within 3 years following lumpectomy. Although a large number of patients have been followed for 5 years, it has been noted that survival thus far appears unrelated to local recurrence in the breast and that the occurrence of this latter is significantly decreased in those women

THE IMPACT OF PATHOLOGY ON BREAST CANCER

1083

receiving postoperative irradiation. Of course much observation will be required to finally assess the biologic and clinical significance of true multicentricity. Results of our recent investigation concerning the incidence and significance of bilateral breast cancer in patients enrolled in Protocol 4 of the NSABP are relevant to the polemic of multicentric cancer. 37 Here again, invasive diagnostic and even therapeutic procedures have been recommended because of a relatively high incidence of noninvasive cancers in the contralateral breasts of women presenting with cancer. Yet, the estimated incidence of such neoplastic change, cited to be as high as 16 per cent in contralateral mastectomy specimens, far exceeds the actual incidence of clinical contralateral cancer, which in our material was 3. 7 per cent invasive and 0.5 per cent noninvasive forms. As indicated above, the incidence of bilateral cancer, as well as the annual accrual of such lesions per patients at risk (which was found to be less than 1 per cent), is low. Much confusion has arisen from categorization of bilateral breast cancer as being metachronous or synchronous. In fact, the vast majority of cases are probably synchronous, as they appear within 5 years following recognition of the ipsilateral cancer. Kinetic considerations indicate the likelihood of the presence of the contralateral cancer at the time the ipsilateral lesion was recognized, even though it was not clinically detectable. Interestingly, despite the mandated clinical surveillance of the contralateral breast in the protocol, the average size of the second cancers was 2.4 em, or only 1 em less than that of the ipsilateral lesion. This reaffirms the difficulty in clinically recognizing small or "early" breast cancers. Most important, the development of a second cancer was not found to influence the expected survival rate of the first or ipsilateral tumor. The occurrence of bilateral breast cancers in women with ipsilateral disease was found to be significantly related to first cancers measuring greater than 2.0 em, and to those with invasive multicentric cancer, nipple involvement, absent nodal sinus histiocytosis, and cancer that was of the lobular invasive and tubular type. However, the degree of risk for these discriminants was no greater than 2 to 3:1. Although a family history was found to be one and one-half to twice as more frequent in patients with bilateral than unilateral breast cancer, this estimate was not statistically significant. The preceding strongly indicates to us that it is justifiable to follow the contralateral breast of patients by careful physical and mammographic examinations rather than contralateral biopsies or mastectomies. This conclusion also appears to support the rationale for lumpectomy in the primary surgical treatment of some breast cancers.

PROGNOSIS The material utilized for the search of prognostic pathologic discriminants has been derived from Protocol 4 of the NSABP. This protocol was ideal for such estimates, as well as other information relating to the natural history of breast cancers, as it was prospectively designed and survival was found to be comparable in all subsets of its two major arms (clinically negative and clinically positive regional lymph node status). Thirty-eight

1084

EDWIN

R.

FISHER

pathologic and six clinical features including survival encountered in 1665 patients with invasive breast cancer were examined by appropriate statistical methods. Preliminary identification of prognostic discriminants disclosed 20 features that might be considered related to prognosis. Ranking these variables consistently revealed the pathologic nodal status to be the most dominant influence on survival, whether treatment failure was assessed for 2, 5, or 10 years postoperatively. In addition, the number of involved nodes was prognostically discriminating; patients with one to three nodal metastases fared better than those with involvement of four or more. More recently it has been demonstrated that when this latter group is subdivided, treatment failure is increasingly more frequent in those women with 4 to 7, 7 to 12, and 13 or more nodal metastases. 14 Although prognosis may be crudely related to the clinical assessment of the axillary nodal status, it must be emphasized that accurate staging of the disease can be obtained only from pathologic examination. There is an approximately 40 per cent false-positive and false-negative error rate inherent in clinical assessment of nodal status. 19 Contrary to the interpretations of some, 5 there does not appear to be any gain in survival in those patients exhibiting a false-positive clinical appraisal. The importance of assessing the number of nodal metastases is reflected by the results of studies relating to their size, level, and extent. No difference in survival has been observed between patients whose nodal metastases measure less than 2 mm in greatest diameter (micrometastases) and those with negative nodes. 24 • 33 The level of nodal involvement has been regarded as an important discriminant for estimating survival. However, Smith and associates 54 found a close correlation between the total number of nodal metastases and the level of such involvement. The extension of metastases through the capsule of involved nodes, an ominous finding, has also been observed to occur more frequently in patients with four or more involved nodes. 22 The failure to recognize occult nodal metastases as a distinct subset of nodal metastases in therapeutic and prognostic considerations might well account for the enigmatic behavior of the cancer in those patients with one to three positive nodal metastases. Perhaps patients in whom such nodal metastases are observed should be regarded as having "stage 1.5" disease. 24 • 26 As estimation of pathologic nodal status is critical to therapeutic and prognostic considerations, should a simple axillary sampling or a more extensive dissection be performed for this purpose? Fisher and associates 13 recently wrote that the qualitative status (that is, whether nodes are positive or negative) might be established accurately from the removal of as few as three to five nodes. Conversely, the quantification of nodal involvement required removal of 10. Thus, the quantitative nodal information required would necessitate an axillary dissection directed toward the removal of nodes from at least levels 1 and 2. Because of the importance of nodal status, biologically as well as prognostically, it was considered most appropriate to attempt to discern features that might provide further discrimination in these various nodal categories. Unfortunately, samples sizes were small in the categories of 4 to 7, 7 to 12, and 13 or more positive nodal metastases, disallowing for evaluations in these groups. The investigations35 were performed only on

THE IMPACT OF PATHOLOGY ON BREAST CANCER

I085

the 654 patients who were subjected to radical mastectomy in accord with the importance of pathologic determination of nodal status indicated above. Table I indicates the actual treatment failure rates observed in the fifth and tenth postmastectomy years of observation according to the various nodal categories. Interestingly, there is no striking difference in treatment failure rates between the 5- and 10-year estimates. 30• 35 Life tables encompassing the first to tenth year of disease-free survival when lymph node status was adjusted or controlled revealed only a tumor size smaller than 2 em to be positively related to survival and independent of nodal status. A definite but not statistically significant trend was noted for a more favorable outcome at 10 years with scar cancer types I, 2 and 4. Yet no characteristics except the presence of 13 or more nodes and occurrence of scar cancers types 3 and 5 were found to be significant in prediction of which patients who were disease-free at 5 years would not exhibit disease-free survival in the tenth year of observation. When the life tables were prepared according to nodal categories (that is, either absent, one to three, and four or more positive nodes; several characteristics were found to be significantly related to disease-free state (Table 2). An adverse effect was significantly noted in patients with negative nodes whose lymph nodes exhibited a germinal center predominance pattern or appearance and whose tumors were assessed as being most malignant (histologic grade 3 by our conventional method). 19· 29 Strong trends of this relationship were also observed with tumors less than 2 em and scar cancer types 3 and 5 or no scar cancers. No significant consistent characteristics were encountered in the group with one to three positive nodes but patients with more than four nodes, whose tumors exhibited the most malignant histologic grade and were larger than 1. 9 em, fared poorly at 10 years. Several schemes have been utilized for the histologic grading of breast cancers. Most, if not all, basically utilize nuclear characteristics encompassed by assessment of nuclear grade and the identification of tubule formation. As a result, we have been utilizing a rather simplified method for histologic grading based upon these two factors (Table 3). It must be emphasized that estimation of nuclear grade and tubule formation is based only on the invasive component of the tumor. The intraductal component is regarded as a distinct entity that is assessed independent of the invasive elements. It might be noted that failure to adhere to this condition unquestionably accounts for much of the confusion, uncertainty, and lack or reproducibility not only the histologic grading of breast cancers, but also of their histologic typing. It should also be noted that the method for estimating nuclear grade represents a modification of the original method described by Black and Speer. 4 The departures from this latter method consist of condensing the original five types of nuclear dedifferentiation into three categories. Also, the order of numerical assessment of nuclear atypia is reversed from that utilized by Black and Speer so that nuclear grade 1 represents the most differentiated and nuclear grade 3 the least differentiated nuclear forms-a practice that more logically conforms to the customarily employed designations of histologic grading and staging of neoplasms. In a previous study, we found no significant difference in disease-free survival in histologic grade 2 cancers stratified according to

~--------

1086

EDWIN

R.

FISHER

Table 1. Nodal Status and Survival PER NODES

CEJ~;T

SURVIVAL

NO. PATIENTS

At 5 Yr

At 10 Yr

279

87

80

160 175 (65) (55) (55)

61 31

53 29 41 31 23

Negative nodes Positive nodes: 1--3

4+ 4--6

7-12 13+

614

Total

Table 2. Pathologic Features Related to 10-Year Treatment Failure According to Nodal Category NODES

PROBABILITY

Negative Nodes Nodal germinal center predominance Histologic grade 3 Tumors 2: 2 em, scar cancer type 3.5

p = 0.004 p = 0.02 Trend

1-3 Positive Nodes None with significance 4 + Positive Nodes Tumors 2: 2 em Histologic grade 3

p p

0.01 0.015

Table 3. NSABP* Conventional Scheme of Histologic Grading for Breast Cancer TUBULES (INCLUDING ADENOCYSTIC FORMS)

Pure, in part Pure, in part Absent Absent *NSABP

=

NUCLEAR GRADE

HISTOLOGIC GRADE

1 2, 3 1 2,3

1 2 2 3

National Surgical Adjuvant Breast Project.

1087

THE IMPACT OF PATHOLOGY ON BREAST CANCER

degree of tubule formation. 29 However, when both histologic grades 1 and 2 were examined, a statistically significant increase in survival was observed to be_ directly related to the detection of a marked degree of tubule formation. The association of a marked tubular component with tumors exhibiting the highest degree of differentiation (histologic grade 1) appears noteworthy. The relationship of marked tubule formation to survival was not found to be dependent on nodal status. The tubular or ductal structures comprising tubular carcinomas or cancers containing some tubular elements are generally regarded as being well differentiated. However, we have long been impressed that less differentiated tubular structures are not infrequent. 15 Their appearance is more complicated or glandular than the simple, angulated forms of the prototype tubular cancer. They may appear as solitary structures or in an adenocystic or cribriform pattern. Linell and associates 45 have also alluded to the occurrence of this second type of tubular structure by noting that some tubuloductal carcinomas progress from tubules to less differentiated ductal structures. Shorter survival time was noted in those patients whose cancers contain the complicated or glandular rather than the simple prototype form of tubules. 33 As with degree of tubules, no consideration is given to the type of tubule formation in our conventional histologic grading system. Attempts to integrate both type and degree of tubule formation into a revised grading scheme were unrewarding and complicated. It would appear that the two criteria are interchangeable. Also, there does not appear to be any salient shortcoming from categorizing breast cancers into two general groups by combining tumors of histologic grades 1 and 2 as representative of welldifferentiated lesions, as opposed to the poorly differentiated or grade 3 types, as we have done on occasion. This practice possesses the advantage of simplicity. The failure to detect any significant discriminant for treatment failure in patients with one to three positive nodes is perplexing and enigmatic. It is relevant, however, to recognize that the size of the nodal metastases in almost 25 per cent of these patients is smaller than 2 mm or even 1.3 mm. Such sizes have been noted to be attendant with survival indistinguishable from that of patients without nodal metastases. This countereffect of a reasonably significant number of patients may account for the failure to detect any statistically significant prognostic discriminants in this group of patients. The prognostic significance of demonstrating intralymphatic and blood vessel invasion in sections of breast cancers has been the subject of recent Table 4. NSABP Alternate Scheme of Histologic Grading for Breast Cancer TUBULES (INCLUDING ADENOCYSTIC FORMS)

Pure or marked (simple) Slight, moderate (glandular complicated) Absent Absent

NUCLEAR GRADE

HISTOLOGIC GRADE

1, 2, 3 1, 2, 3 1 2, 3

2 2

1 3

1088

EDWIN

R.

FISHER

controversy. One of the dilemmas confronting those concerned with the treatment of patients with breast cancer is whether to administer adjuvant chemotherapy to patients with pathologic stage I disease in whom pathologic examination has revealed intralymphatic or blood vessel invasion. Nime and co-workers 50 have reported the incidence of distant metastases to be greater in patients designated as having stage I disease with intralymphatic tumor emboli than in those lacking this finding. Similar observations subsequently have been recorded by Bettelheim and Neville, 3 Nealon and associates, 49 and Roses and associates. 52 Only Bettelheim and Neville and Roses and associates mention whether lymphatics within the vicinity of the dominant mass or in quadrants remote from the dominant mass were involved. This distinction is important, as lymphatic extension into remote breast quadrants represents a histologic manifestation of inflammatory breast cancer, a form of the disease attendant with almost universal treatment failure and other ominous features. It should be recognized that, except for the study of Nealon and co-workers, few patients were examined. A more disquieting concern relative to the role of intralymphatic extension as a discriminant is the lack of unanimity among pathologists concerning the recognition of this phenomenon. Gilchrist and associates 39 have recounted the failure of three pathologists to exhibit significant concordance regarding the detection of lymphatic embolization in tissue sections in a double-blind study of test specimens. It is true that many breast cancers contain aggregates of tumor cells in tissue spaces, which may stimulate intralymphatic extension. However, this effect appears to be related to fixation. Careful observation of these artifactual spaces reveals an absence of lining endothelia. Our experience has taught us to disregard such questionable examples. True lymphatic extension is more clearly perceived at the periphery but still within the immediate vicinity of the dominant mass. With these criteria, at least among our group, we find a reproducibility of 92 per cent in the estimate oflymphatic extension. Despite our confidence in discerning this characteristic, the data obtained indicated that the phenomenon was more closely related to nodal status than treatment failure. 35 Indeed, 33 per cent of the tumors from patients surviving for 10 years possessed such extension. Dawson and associates 6 observed lymphatic extension in 63 per cent of 25-year survivors. It is not surprising to us that lymphatic extension does not appear as a significant discriminant for treatment failure when it is appreciated that the survival of patients in whom nodal micrometastases (smaller than 2 mm) are found is similar to that of patients with negative nodes. Another equally controversial issue is that concerning the prognostic significance of blood vessel invasion. Reports reveal an incidence of such an event to vary from 21 to 46 per cent. Our own estimate of only approximately 4 per cent19 appears to be at marked variance with these figures; however, it is important to note that our studies have relied on the more classical features of tumor thrombus, of tumor covered by endothelial cells in a luminated, vis-a-vis vascular space, or of the presence of neoplastic cells in an unequivocal venule or capillary.

THE IMPACT OF PATHOLOGY ON BREAST CANCER

1089

These criteria have not been universally utilized in studies citing a higher frequency of blood vessel invasion. Some have regarded the presence of distorted and disoriented elastic fibers associated with tumor cell aggregates in the vicinity of the ductal-vascular bundles sufficient to connote blood vessel invasion. 55 In one study, blood vessel invasion in "arteries and veins" was scored positive58 but intra-arterial neoplastic extension is conventionally regarded as a curiosity. We have always contended our estimate of blood vessel extension in breast carcinoma is conservative. Indeed, we have identified actual tumor cells in mammary vein blood in 26 per cent of patients with stage I and stage II disease undergoing modified radical mastectomy. 42 The disparity in this and the hi3topathologic estimates is not surprising, as we have observed a similar difference in a comparison of the incidence of circulating tumor cells and histopathologic demonstration of vascular extension in colorectal carcinoma. 18 Certainly, the biologic significance of such circulating tumors cells in breast carcinoma is not known and, if comparable to the experience with colorectal carcinoma, may not be prognostically significant. 7 It should also be remembered that only about 50 per cent of patients with angioinvasive thyroidal carcinoma will die of their disease, although diagnosis is based on the presence of blood vessel invasion. This information reaffirms Goldmann' s41 1897 dictum that not all circulating tumor cells necessarily cause metastases. Interestingly, assessment of our material for blood vessel invasion has been performed by another pathologist who utilizes what we regard as less stringent criteria for such an event. 55 However, no statistical relationship between his assessment and 10-year survival was found. Similarly, he could find no relation between survival and his estimates of lymphatic invasion in our material. RECEPTORS: CHEMOTHERAPY In previous pathologic studies, we have observed a significantly greater number of well-differentiated breast cancers in older (> 54 years) and postmenopausal than younger (< 49 years) or premenopausal women. 19 This finding correlated with the well-recognized observation of a greater incidence of estrogen receptor (ER) positive cancers in older women. Indeed, a strong association between ER status and degree of tumor differentiation has been observed. 31 • 32 This information provoked the view that the chemotherapeutic unresponsiveness purported to occur in older patients by some might be related to the biologic nature, vis-a-vis histologic differentiation, of their neoplasm. A relationship between tumor responsiveness to radiographic therapy (and also perhaps radiomimetic agents) and tumor differentiation has long been recognized as dictum. That this perception might be valid in regard to chemotherapeutic responsiveness appears to be substantiated by our recent studies. It has been noted that regimens of L-PAM, L-PAM + 5FU (fluorouracil), and Lp AM + 5FU + methotrexate in NSABP clinical trials effected an increased survival compared with controls not receiving such adjuvant therapy. This

1090

EDWIN

R.

FISHER

favorable response was more pronounced in those women whose cancers were assessed pathologically to be poorly differentiated (histologic grade 3) and exhibited four or more nodal metastases, regardless of age, than in those with well-differentiated cancers and similar nodal status. Indeed, a clear advantage for the use of adjuvant chemotherapy was evident for those women 50 years of age or more whose tumors were poorly differentiated. Interestingly, Millis and associates 48 have noted a more favorable response to endocrine therapy in those patients with advanced breast cancer whose tumors were well differentiated rather than poorly differentiated. These findings indicate the importance of pathologic assessment for identifYing subsets of patients not only from a biologic or prognostic perspective, but also from a therapeutic one.

FROZEN SECTION: THE ONE-STEP PROCEDURE Rapid diagnosis by frozen section is a valuable methodology in the armamentarium of the surgical pathologist. However, its utilization by surgeons might be regarded in many instances as more ritualistic than realistic. There is unquestionably a place for its use in some so-called onestep procedures performed on the breast. However, with the advent of alternate forms of primary surgical treatment of breast cancer, it should become apparent that not only clinical but also pathologic input into the decision-making process will be essential. The necessary pathologic assessments cannot be performed on sections prepared by the rapid techniques available nor on the type and quantity of tissue often submitted for such an evaluation. There has been a striking increase in the demand for rapid diagnosis of very small (< 0.5 em) breast lesions due to the widespread use of mammography. It is often difficult to recover sufficient tissue for proper rapid diagnosis, let alone diagnosis based on conventional techniques and the biochemical assessment of receptor status. Certainly, therapeutic decisions based on the rapid diagnosis of intraductal hyperplastic lesions is extremely hazardous. This difficulty should be appreciated when it is realized that many such lesions represent a stage in the evolution of cancer. Indeed, most errors in frozen section diagnosis appear attributable to misinterpretation of such lesions that are even "worrisome" in the less rapid conventionally prepared sections. It is becoming increasingly apparent that breast cancer is not an emergency disease, a view that apparently has its roots in experiments performed at the turn of the century in Tyzzer, 56 who purportedly demonstrated metastases to follow tumor manipulation in the Japanese waltzing mouse. From this the concept evolved that mechanical trauma dislodged tumor cells and any subsequent delay would provide time for such dislodged cells to reach regional lymph nodes. Not only has there never been any significant confirmation of Tyzzer' s observations, but it has not been demonstrated that the actual presence of tumor cells within the regional venous system draining some cancers actually influences survival, as noted previously. Also, it is highly unlikely that a surgeon could perform the mastectomy quickly enough to forestall such spread, since tumor cells have I

THE IMPACT OF PATHOLOGY ON BREAST CANCER

1091

been demonstrated experimentally to not only traverse nodes but also interstitial spaces with alarming alacrity. Most clinical studies purporting to declare a therapeutic advantage for the one-step procedure exhibit crucial statistical deficiencies, as noted by Margolese47 in his review. Preliminary appraisal of our own NSABP Protocol 4 material in this regard revealed no statistically significant advantage in 10-year survival rates for women undergoing a one-step procedure.

REFERENCES l. Andersen, J.A., and Schiodt, T.: On the concept of carcinoma in situ of the breast. Pathol. Res. Pract. 166:407--414, 1980. 2. Betsill, W. L., Rosen, P. P., Lierberman, P. H., eta!.: Intraductal carcinoma: Longterm follow-up after treatment by biopsy alone. J.A.M.A., 239:1863-1867, 1978. 3. Bettleheim, R., and Neville, A. M.: Lymphatic and vascular channel involvement within infiltrative breast carcinomas as a guide to prognosis at the time of primary surgical treatment. Lancet, 2:631, 1981. 4. Black, M. D., and Speer, F. S.: Nuclear structure in cancer tissues. Surg. Gynecol. Obstet., 105:97-102, 1957. 5. Cutler, S. J., Black, M. M., and Goldenberg, I. S.: Prognostic factors in cancer of the female breast. Cancer, 16:1589-1597, 1963. 6. Dawson, P. J., Ferguson, D. J., and Karrison, T.: The pathologic findings of breast cancer in patients surviving 25 years after radical mastectomy. Cancer, 50:2131-2138, 1982. 7. Engel!, H. D.: Cancer cells in the circulating blood. Acta Chir. Scand. [Suppl.], 201:1-70,

1955. 8. Fisher, B.: Laboratory and clinical research in breast cancer-Personal adventure. The David A Karnofsky Lecture. Cancer, 40:3863-3874, 1980. 9. Fisher, B., and Fisher, E. R.: Transmigration of lymph nodes by tumor cells. Science, 152:1397-1398, 1966. 10. Fisher, B., and Fisher, E. R.: Barrier function of lymph node to tumor cells and erythrocytes. I. Normal nodes. Cancer, 20:1907-1913, 1967. 11. Fisher, B., and Fisher, E. R.: Barrier function of lymph node to tumor cells and erythrocytes. II. Effect of x-ray, inflammation, sensitization and tumor growth. Cancer, 20:1914-1919, 1967. 12. Fisher, B., Montague, E., Redmond, C., eta!.: Comparison of radical mastectomy with alternative treatments for primary breast cancer: A first report of results from a prospective randomized clinical trial. Cancer, 39:2827, 1977. 13. Fisher, B., Wolmark, N., Bauer, M., eta!.: The accuracy of clinical nodal staging and of limited axillary dissection as a determinant of histologic nodal status in carcinoma of the breast. Surg. Gynecol. Obstet., 152:76!>-772, 1981. 14. Fisher, B., Bauer, M., Wickerham, D. L., et a!.: Relationship of number of positive axillary nodes to the prognosis of patients with primary breast cancer-An NSABP update. Cancer, 52:1551-1557, 1983. 15. Fisher, E. R.: Ultrastructure of human breast and its disorders. Am. J. Clin. Pathol., 66:291-374, 1976. 16. Fisher, E. R.: Relationship of fibrocystic disease to cancer of the breast. In Hoogstraten, B., and McDivitt, R. W. (eds.): Breast Cancer. Boca Raton, FL, CRC Press, 1981. 17. Fisher, E. R.: The pathology of breast cancer as it relates to its evolution, prognosis and treatment. In Baum, M. (ed.): Breast Cancer. W. B. Saunders Co., London, 1:703-734, 1982. 18. Fisher, E. R., and Turnbull, R. B.: The cytologic demonstration and significance of tumor cells in the mesenteric venous blood in patients with colo-rectal carcinoma. Surg. Gynecol. Obstet., 100:102-108, 1955. 19. Fisher, E. R., Gregorio, R., and Fisher, B.: The pathology of invasive breast cancer. A syllabus derived from the findings of the National Surgical Adjuvant Breast Project (Protocol 4). Cancer, 36:1-85, 1975. 20. Fisher, E. R., Shoemaker, R. H., and Palekar, A. S.: Identification of premalignant

1092

21. 22.

23.

24.

25. 26. 27. 28.

29. 30.

31. 32.

33.

34.

35.

36. 37. 38. 39.

40.

41. 42. 43.

EDWIN

R.

FISHER

hyperplasia in methyl-cholanthrene-induced mammary tumorogenesis. Lab. Invest., 33:446-450, 1975. Fisher, E. R., Shoemaker, R. H., and Sabnis, A.: Relationship of hyperplasia to cancer in 3-methyl-cholanthrene-induced mammary tumorogenesis. Lab. Invest., 33:33-42, 1975. Fisher, E. R., Gregorio, R. M., Redmond, C., et al.: Pathologic findings from the National Surgical Adjuvant Breast Project (Protocol4). III. The significance of extranodal extension of axillary metastases. Am. J. Clin. Pathol., 65:439, 1976. Fisher, E. R., Redmond, C., and Fisher, B.: A perspective concerning the relationship of duration of symptoms to treatment failure in patients with breast cancer. Cancer, 40:3160-3167, 1977. Fisher, E. R., Palekar, A. S., Rockette, H., et al.: Pathologic findings from the National Surgical Adjuvant Breast Project (Protocol 4). V. Significance of axillary nodal micro and macrometastases. Cancer, 42:2032-2038, 1978. Fisher, E. R., and Paulson, J.D.: Karyotypic abnormalities in precursor lesions of human cancer of breast. Am. J. Clin. Pathol., 69:284-288, 1978. Fisher, E. R., Swamidoss, S., Lee, C. H., et al.: Detection and significance of occult axillary metastases in patients with invasive breast cancer. Cancer, 42:2025-2031, 1978. Fisher, E. R., Palekar, A. S., Kotwal, N., eta!.: A non-encapsulated sclerosing lesion of the breast. Am. J. Clin. Pathol., 71:240-246, 1979. Fisher, E. R., Palekar, A. S., Redmond, C., et al.: Pathologic findings from the National Surgical Adjuvant Breast Project. Invasive papillary cancer. Am. J. Clin. · Pathol., 73:313-322, 1980. Fisher, E. R., Redmond, C., and Fisher, B.: Histologic grading of breast cancer. Pathol. Annu., 15:239-251, 1980. Fisher, E. R., Redmond, C., and Fisher, B.: Pathologic findings from the National Surgical Adjuvant Breast Project (Protocol 4). VI. Discriminants for five-year treatment failure. Cancer, 46:908-918, 1980. Fisher, E. R., Redmond, C., Liu, H., et al.: Correlation of estrogen receptor and pathologic characteristics of invasive breast cancer. Cancer, 45:349-353, 1980. Fisher, E. R., Osborne, C. K., McGuire, W. L., et al.: Correlation of primary breast cancer histopathology and estrogen receptor content. Breast Cancer Res. Treat. 1:37-42, 1981. Fisher, E. R., Palekar, A. S., et al.: Scar cancers: Pathologic findings from the National Surgical Adjuvant Breast Project (Protocol 4). IX. Breast Cancer Res. Treat. 3:39-59, 1983. Fisher, E. R., Redmond, C., Fisher, B.: Pathologic findings from the National Surgical Adjuvant Breast Project. VIII. Relationship of chemotherapeutic responsiveness to tumor differentiation. Cancer, 51:181-191, 1983. Fisher, E. R., Sass, R., and Fisher, B.: Pathologic findings from the National Surgical Adjuvant Breast Project (Protocol 4). X. Discriminants for tenth year treatment failure. Cancer, 53:712-723, 1984. Fisher, E. R., and Brown, R.: Intraductal signet ring cells-An undescribed form of intraductal carcinoma of breast. Am. J. Surg. Pathol. (in press). Fisher, E. R., Fisher, B., Sass, R., et al.: Pathologic findings from the National Surgical Adjuvant Breast Project (Protocol 4). XI. Bilateral breast carlcer. Cancer (in press). Gallager, H. S., and Martin, J. E.: An orientation to the concept of minimal breast cancer. Cancer, 28:1505-1507, 1971. Gilchrist, K. W., Gould, V. E., Hirschi, S., et al.: Interobserver variation in the identification of breast carcinoma in intramammary lymphatics. Human Pathol., 13:170-172, 1982. Gimbrone, M. A., Jr., and Gullino, P.: Neovascularization induced by intraocular xenografts of normal, preneoplastic mouse mammary tumor. J. Nat. Cancer lnst., 56:305-318. Goldmann, E. E.: Anatomische Untersuchungen uber die Verbreitungswege bosartiger Geschwulste. Beitr. z. klin. Chir., 18:595, 1897. Golinger, R. C., Gregorio, R., and Fisher, E. R.: Tumor cells in venous blood draining mammary carcinomas. Arch. Surg., 112:707-798, 1977. Gullino, P. M.: Natural history of breast cancer. Progression from hyperplasia to neoplasia as predicted by angiogenesis. Cancer, 39:2697-2703, 1977.

THE IMPACT OF PATHOLOGY ON BREAST CANCER

1093

44. Kramer, W. M., and Rush, B. F.: Mammary duct proliferation in the elderly: A histopathic study. Cancer, 31:130--137, 1973. 45. Linell, F., Ljungsberg, 0., and Andersson, 1.: Breast carcinoma. Aspects of early, stages progression and related problems. Acta Pathol. Microbiol. Scand. [Suppl.] 272, 1980. 46. Mackay, E. N., and Sellars, A. H.: Breast cancer at the Ontario Cancer Clinics, A statistical review. Ottawa, Medical Statistics Branch, Ontario Department of Health, 1938--1956, 1965. 47. Margolese, R. G.: The case for the two step biopsy procedure for breast cancer. CA, 32:51-57, 1983. 48. Millis, R. R., Rubens, R. D., Masters, J. R. W., eta!.: Histologic grade and response to endocrine therapy in breast cancer. Lancet, 2:191, 1981. 49. Nealon, T. R., Nkongho, A., Grossi, C. E., et a!.: Treatment ot early cancer of the breast (T 1N 0 M 0) on the basis of histopathologic characteristics. Surgery, 89:279--289, 1981. 50. Nime, F. A., Rosen, P. P., Thaler, H. T., et a!.: Prognostic significance of tumor emboli in intramammary lymphatics in patients with mammary carcinoma. Am. J. Surg. Pathol., 1:25--30, 1977. 51. Rosen, P. P.: Lobular carcinoma in situ: Recent clinicopathologic studies at Memorial Hospital. Path. Res. Pract., 166:430--455, 1980. 52. Roses, D. R., Bell, D. A., Flotte, T. J., et a!.: Pathologic predictors of recurrence in stage 1 (T1 N0 M0) breast cancer. Am. J. Clin. Pathol., 78:317--320, 1982. 53. Shapiro, S.: Evidence on screening for breast cancer from a randomized trial. Cancer, 39:2772-2782, 1977. 54. Smith, J. A., III, Gamez-Araujo, J. J., Gallager, H. S., eta!.: Carcinoma of the breast. Analysis of total lymph node involvement versus level metastasis. Cancer, 39:527-532, 1977. 55. Somers, S. C.: Personal communication, 1984. 56. Tyzzer, E. E.: Factors in the production and growth of tumor metastases. J. Med. Res., 28:309--333, 1913. 57. Veronesi, U., Saccozzi, R., DelVecchio, M., eta!.: Comparing radical mastectomy with quadrantectomy, axillary dissection, and radiotherapy in patients with small cancers of the breast. N. Engl. J. Med., 305:6-11, 1981. 58. Weigand, R. A., Isenberg, W. M., Russo, J., eta!.: Blood vessel invasion and axillary lymph node involvement as prognostic indicators for human breast cancer. Cancer, 50:962-969, 1982. 59. Zelen, M.: Theory of early detection of breast cancer in the general population. In Heuson, J. C., Mattheiem, W. H., Rozenweig, M. (eds.): Breast Cancer: Trends in Research and Treatment. New York, Raven Press, 1976. Institute of Pathology Shadyside Hospital 5230 Centre Avenue Pittsburgh, Pennsylvania 15232