1990, The British Journal of Radiology, 63, 845-849
The relationship of "high risk" mammographic patterns to histological risk factors for development of cancer in the human breast By Justine E. Arthur, MB, BS, I. O. Ellis, MB, BS, MRCPath, *C. Flowers, MRCS, LRCP, FRCR, *E. Roebuck, DMRD, FRCR, C. W. Elston, MD, MRCPath and tR. W. Blarney, MD, FRCS Departments of Histopathology and tSurgery, City Hospital and *Department of Radiology, University Hospital, Nottingham {Received February 1990 and in revised form May 1990)
Abstract. In the UK Trial for the Early Detection of Breast Cancer in Nottingham, 119 women were identified as having fibrocystic change with epithelial hyperplasia or in situ carcinoma. Their mammograms were classified according to Wolfe's criteria and the corresponding histology for each patient was classified for degrees of epithelial hyperplasia, atypia and in situ neoplasia using Page's criteria. A control population of patients presenting for breast screening was used to represent the general population. No correlation was found between the four mammographic Wolfe patterns, N l , PI, P2 and DY and histological evidence of epithelial hyperplasia, atypia or in situ carcinoma. A further study was carried out to determine histological features of Wolfe pattern, using radiological examination of resected breast tissue. The variation in Wolfe pattern was related to the distribution of fibrous and adipose tissue in the breast interlobular stroma and appeared to have no relationship to epithelial parenchymal content. This information does not support the hypothesis that radiographic densities of P2 and DY patterns correspond to high risk epithelial proliferation.
Each year 15 000 women in England and Wales die from breast cancer and 1 in 15 women will develop the disease in their lifetime (OPCS, 1983, HMSO, 1983). The current lack of any prospect for primary prevention increases the need to develop more effective and less expensive techniques for early detection of breast cancer. Selection of women at high risk might reduce the number who need to attend for regular mammographic screening. The work of Wolfe shows that women exhibiting certain mammographic breast parenchymal patterns are at greater risk of developing breast cancer (Wolfe, 1976, 1987; Wolfe et al, 1983). He has described the following breast parenchymal patterns: Nl the breast is composed primarily of fat, often with a fine trabeculated appearance; PI "prominent" ducts occupy 25% or less of the volume of the breast; P2 "prominent" ducts occupy more than 25% of the volume of the breast; DY mammary dysplasia (fibrocystic change) with sheet-like areas of increased density. Wolfe states that the mammographic patterns most often related to breast cancer and premalignant epithelial alterations, especially after the age of 50, are the P2 and DY breast patterns, Nl and PI being rarely associated with these conditions. He reports that the incidence of subsequent carcinoma development, per 1000 women, over a period of 3 years increases 30 fold, from Address for correspondence: Dr I. O. Ellis, Department of Histolopathology, City Hospital, Hucknall Road, Nottingham NG5 1PB. Vol. 63, No. 755
1.45 cases per 1000 women with Nl patterns to 44.60 cases per 1000 women with DY patterns. (PI = 3.85 and P2 = 17.45) Several categories of epithelial proliferation have been described in the female breast (Page & Anderson, 1988): regular epithelial hyperplasia of mild, moderate and florid degree, atypical ductal hyperplasia and atypical lobular hyperplasia; and ductal and lobular carcinoma in situ. Increasing risks of subsequent development of breast carcinoma have been demonstrated with these conditions (Page et al, 1978, 1982, 1985). Mild hyperplasia carries the same risk as the general population of the subsequent development of invasive breast carcinoma. Moderate and florid hyperplasia carry a risk double that of the general population for cancer development. Atypical hyperplasia of ductal and lobular types are associated with four times the risk for subsequent development of invasive carcinoma. Ductal and lobular carcinomas in situ are associated with a risk for invasive carcinoma development 10 times that of the general population (Hutter et al, 1986; Page et al, 1978, 1982; Page & Anderson, 1988). Despite these detailed studies of histopathological and radiological risks for carcinoma development in the breast, little work has been undertaken to relate these phenomena to each other. One study, performed by Wellings and Wolfe (1978) correlated the radiographic densities of PI, P2 and DY mammographic patterns withfibrosisand a variety of lesions of the terminal duct lobular units particularly "atypical lobules Type A". Two of these lesions were described previously by Wellings and Jensen (1975) but both were thought to arise from a similar histogenetic pathway. Type A 845
Justine E. Arthur et al Table I. The study population identified from the Nottingham Breast Self Examination (BSE) arm of the UK Trial for the Early Detection of Breast Cancer 1979-1986
Method
Between 1979 and 1986, 56000 women aged 45-64 years in Nottingham were invited for education in Breast Self Examination as part of the United Kingdom Women attending for Trial of Early Detection of Breast Cancer. Women BSE 28 759 (49.86% of those invited) participating in this programme were offered an open Number of biopsies in access mammography and clinical examintion service the study group for should they detect a breast abnormality through self malignant conditions 546 ~) examination. If a significant lesion was identified at Number of biopsies in >119 borderline biopsies* the study group for clinic attendance, open biopsy was performed. During benign conditions 449J this study period biopsies were carried out on 995 of the study group. One hundred and nineteen cases were * Borderline biopsies indicates the number of biopsies for identified as "borderline lesions" of epithelial hypercarcinoma in situ (included in the malignant category) and plasia and in situ carcinoma (Table I). epithelial hyperplasia (included in the benign category). In our study, these 119 patients with borderline lesions were taken to represent women with histological evidence of increased risk. Forty-three women who lesions represent the spectrum of ductal hyperplasia, underwent biopsies which showed benign non-proliferaatypical ductal hyperplasia and ductal carcinoma in situ. tive fibrocystic change are taken to represent women Type B lesions represent the lobular equivalents, that is, with no histological evidence of increased risk atypical lobular hyperplasia and lobular carcinoma (Table II). To represent the general population, 91 in situ. Wellings and Wolfe (1978) found that the degree women, aged between 50 and 64, consecutively of fibrosis and the grade of Type A atypical lobules attending for mammographic breast screening at the increased from PI to DY. A recent study by Urbanski et Nottingham City Hospital during 1998, were used to al (1988) demonstrated a weak relationship between the determine the frequency of each mammographic extent of mammographic "dysplasia" and histological pattern. All mammograms on the patients were availevidence of epithelial atypia and carcinoma in situ in able. The Wolfe classification of all these mammograms women younger than 50 years old. Such findings could was reviewed by a single radiologist to elimiate interexplain the increased risk for cancer development asso- observer variation in the objective grading of mammociated with P2 and DY breasts. grams. The radiological classification based on the We have investigated this putative relationship mammographic pattern was then compared with the between Wolfe pattern and epithelial proliferation with histological findings. Improvement of mammographic two studies. First, a retrospective study was carried out technique has occurred between 1979 and 1988. of the mammographic appearances and histology of However, the differences was not considered sufficient to women from the Nottingham arm of the UK Trial for alter interpretation of the Wolfe pattern. the Early Detection of Breast Cancer. Secondly a The histological slides were stained in Ehrlich's prospective study was performed combining radiolo- haematoxylin and eosin (H&E), and were assessed by gical and histological examination of mastectomy speci- two pathologists using Page's method of classification mens from women undergoing treatment for breast (Page & Anderson, 1988). Mild regular hyperplasia is carcinoma. defined by the recognition of between two and four
Table II. Distribution of the cases according to Wolfe mammographic pattern categories (percentages in parentheses) Screened population
FCC
Nl PI P2 DY
15 12 16 48
(16) (13) (18) (53)
1 (2) 9 (21) 15 (35) 18 (42)
Total
91 (100)
43 (100)
All pathological "borderline" groups
Pathological "borderline" groups Mild EH
Mod EH
(11) (20) (27) (42)
0 (0) 2 (12) 3 (19) 11 (69)
3 (10) 8 (28) 9 (31) 9 (31)
119 (100)
16 (100)
29 (100)
13 24 32 50
ADH 0 0 2 3
(0) (0) (40) (60)
5 (100)
DCIS 8 12 15 23
(14) (21) (26) (39)
58 (100)
ALH 0 (0) 0 (0) 0 (0) 0 (0) 0
(0)
LCIS 2 2 3 4
(18) (18) (27) (36)
11 (100)
FCC = fibrocystic change. Mild EH = mild epithelial hyperplasia. Mod EH = moderate and florid epithelial hyperplasia. ADH = atypical ductal hyperplasia. DCIS = ductal carcinoma in situ. ALH = atypical lobular hyperplasia. LCIS = lobular carcinoma in situ.
846
The British Journal of Radiology, November 1990
"High risk" mammographic patterns and histological risk factors in breast cancer ] _„
Mammographic Parenchymal •
80
Pattern
60-
50-
40-
30-
10-
30-
II A
B
CD
E
F
N1 P1 P2 Mammographic Parenchymal Pattern
Histological Risk Group
Figure 1. The distribution of Wolfe mammographic breast parenchymal patterns in various histological subgroups. A: Screened/general population; B: non-proliferative fibrocystic change; C: mild regular epithelial hyperplasia; D: moderate florid regular epithelial hyperplasia, mild atypical hyperplasia; E: atypical epithelial hyperplasia; F: carcinoma in situ (ductal and lobular).
epithelial cells above the basement membrane. Moderate and florid regular hyperplasia are recognized as five or more cells above the basement membrane, with a tendency to distension and crossing of the space by hyperplastic cells which do not show the cytological and architectural patterns of atypical hyperplasia and ductal carcinoma in situ. Atypical ductal hyperplasia is recognized by its histological pattern which is intermediate between that of regular hyperplasia and ductal carcinoma in situ, that is, with some but not all the histological features of ductal carcinoma in situ. Ductal carcinoma in situ in its various subtypes is recognized by its characteristic architectural and cytological features, well described by Page and Anderson (1988). Atypical lobular hyperplasia and lobular carcinoma in situ are recognized by an expansion of small uniform cells with lobular units which may extend to involve ducts in a Pagetoid manner. Filling of all acini within the lobule with distortion of at least 50% of the lobule is used as the arbitrary divide between the two conditions. An additional prospective study was also performed involving a radiographic and histological study of breast slices from 17 consecutive women who had undergone mastectomy. All of these patients had had pre-operative mammography. It is not possible to examine the entire breast in a histologicaJ section. Specimen radiography was used to confirm that the areas of breast sampled had a radiographic appearance representative of the mammographic pattern. Slices measuring 1-2 cm in Vol. 63, No. 755
Figure 2. The distribution of the normal population and histological risk groups found in each of the Wolfe mammographic parenchymal patterns. A-E: as in Fig. 1; F: ductal carcinoma in situ; G: lobular carcinoma in situ.
thickness were taken from each mastectomy specimen. The slice included the nipple and was taken from as near as possible to the 3-9 o'clock position in the horizontal plane. A specimen radiograph was taken and correlated with the pre-operative cranio-caudal mammographic view. Using the specimen radiograph and mammogram it was possible to sample areas of tissue containing the radiographic densities characteristic of the mammographic pattern. On average, 10 blocks were taken from each slice, and between four and six sections were cut and stained in Ehrlich's H & E. The direction and position of each block was recorded on diagrams corresponding with the mammograms and specimen radiographs. Results
The distribution of breast patterns seen in the control group and in each histological group is shown in Table II and illustrated in Fig. 1. There was no correlation between the radiological and the histological risk factors for development of breast cancer. Figure 2 shows the overall distribution of the histological conditions seen in each mammographic subgroup. These results demonstrate that any increased risk associated to P2 and DY patterns cannot be attributed to a greater proportion of high risk epithelial abnormalities in the breasts. From the study of specimen radiology of women who had undergone mastectomy, it appears that the variation in Wolfe pattern was a result of the ratio of fibrous to adipose tissue in the breast with no apparent relationship to epithelial parenchymal content. A varying distribution of collagenous fibrous tissue and adipose tissue 847
Justine E. Arthur et al
was observed in the breasts examined. A classification of this distribution based on the arrangement of these tissues was devised. Three basic patterns were observed: (1) septae, where fibrous tissue was confined to thin fibrous septae usually containing vessels and subsegmental ducts; (2) islands, where the fibrous tissue was arranged in nodular collections, usually containing parenchymal structures, and surrounded by adipose tissue; (3) confluent, where the fibrous tissue was arranged as a diffuse sheet with little adipose tissue present. Within some breasts, more than one type of distribution was observed across the breast disc. The dominant pattern was recorded in these instances. The results of this aspect of the study are shown in Table III. In addition, the proportion of epithelial parenchymal structures present was recorded. This was highly variable between cases with no apparent relationship to mammographic background pattern or histological fibrous tissue and adipose tissue distribution. Discussion We have examined the putative link between mammographic breast parenchymal pattern and histological evidence of epithelial proliferation, a documented risk factor for the development of breast cancer. If the high risk associated with P2 and DY Wolfe patterns is to be attributed to high grades of epithelial premalignant abnormality, one would expect the proportion of these high risk mammographic patterns to be increased in the high risk histological groups. No such distribution is seen in this study and our findings do not support the results reported by Wellings and Wolfe (1978). In fact, the ratio of high risk (P2 and DY) to low risk (Nl and PI) mammographic patterns is increased in the low risk histology control group and screening control group compared with in the high risk histological groups of in situ carcinoma. In all the histological groups, and in the two control populations, the commonest mammographic categories were P2 and DY (Figs 1 and 2). This observation, although well recognized by radiologists, is not well documented. It could provide an alternative explanation why an increased incidence of carcinoma was demonstrated with these two mammographic patterns. The failure of this study to confirm a correlation between Wolfe mammographic parenchymal pattern and histological epithelial alteration suggests that any high risk for carcinoma development associated with P2 and DY patterns is not due to an epithelial abnormality. Instead, the variation in mammographic patterns appears to be related to the variable ratio and distribution of fibrous tissue and adipose tissue in the breast interlobular stroma. Nl patterns reflect a stroma predominantly composed of adipose tissue with only thin fibrous tissue septae being present. The PI and P2 patterns result from stroma composed of islands and septae of fibrous tissue within adipose tissue. In DY patterns there is coalescence of extensive fibrous islands and septae to produce large areas of virtual confluence of fibrous connective tissue (Table III). 848
Table III. Histological distribution of fibrous tissue within Wolfe radiological groups Breast Pattern
Septae
Islands*
Confluent*
Nl (n = 1) PI (« = 1) P2 (n = 8) DY (« = 7)
1 0 4 3
0 1 8 4
0 0 2 7
* Variable epithelial parenchymal content.
These descriptions of the association of high risk mammographic patterns with the distribution of fibrous tissue are consistent with the findings of Fisher et al (1978) and Mansel (personal communication), but are contrary to prevailing views that associate the radiographic densities of P2 and DY patterns and their corresponding risk for breast cancer development with proliferative fibrocystic change (Wellings & Wolfe, 1978). Although Urbanski et al (1988) have demonstrated a significant association between epithelial atypia and carcinoma in situ and mammographic "dysplasia", this associaton was very weak and could not be reproduced on independent review by a second radiologist. They recognized that the epithelial changes observed were not directly responsible for the mammographic findings. We are unable in this study to address directly the association between subsequent risk of breast cancer and mammographic background pattern. Our observation that these radiographic densities, which are characteristic of the four mammographic patterns, are simply a result of variation in fibrous tissue in the interlobular stroma does not necessarily dispel claims of risk associated with P2 and DY patterns. To our knowledge there are no studies which have investigated the distribution of stromal components to risk or development of breast cancer. Further evaluation of the radiographic determinants of risk and these stromal alterations in the development of breast carcinoma is required. Acknowledgments We wish to thank the administrative staff of the Nottingham Breast Screening Service for their help in identification of the cases studied, the audiovisual department for assistance in preparing the figures and Julie Rainbow and Jinny Spence for preparing the manuscript. References FISHER, E. R, PALEKAR, A., KIM, W. S. & REDMOND, C ,
1978.
The histopathology of mammographic patterns. American Journal of Clinical Pathology, 69, 241-426. HMSO, 1983. Cancer Statistic Registrations (HMSO, London). HUTTER, R. P. V., 1986. Consensus meeting: Is 'fibrocystic disease' of the breast precancerous? Archives of Pathology and Laboratory Medicine, 110, 171-173. OPCS, 1983. Cancer Statistics Registrations (OPCS, England and Wales). PAGE,
D. L.
&
Histopathology Edinburgh).
ANDERSON,
of
the
Breast
T. J.,
1988.
(Churchill
Diagnostic
Livingstone,
The British Journal of Radiology, November 1990
"High risk" mammographic patterns and histological risk factors in breast cancer PAGE, D. L., ZWAGG, R. V., ROGERS, L. W., WILLIAMS, L. T., WALKER, W. E. & HARTMANN, W. H., 1978. Relation
precancerous lesions. Journal of the National Cancer Institute, 55, 231-273.
between the component parts of nbrocystic disease complex and breast cancer. Journal of the National Cancer Institute, 61, 1055-1063.
WELLINGS, S. R. & WOLFE, J. N., 1978. Correlative studies of
PAGE,
D. L.,
DUPONT,
W. D.,
ROGERS,
L. W.
&
LANDENBERGER, M., 1982. Intraductal carcinoma of the breast: follow-up after biopsy. Cancer, 49, 751-758. PAGE, D. L., DUPONT, W. D., ROGERS, L. W., WILLIAMS, L. T., WALKER, W. E. & HARTMANN, W. H., 1985. Atypical
hyperplastic lesions of the female breast—long term follow study. Cancer, 55, 2698-2708. URBANSKI, S., JENSEN, H. M., COOKE, G., MCFARLANE, D., SHANNON, P., KRUIKOV, V. & BOYD, N. F., 1988. The
association of histological and radiological indicators of breast cancer risk. British Journal of Cancer, 58, 474-479. WELLINGS, S. R. & JENSEN, H. M., 1979. An atlas of subgross
the histological and radiographic appearances of the breast with special reference to possible precancerous lesions. Diagnostic Radiology, 129, 299-306. WOLFE, J. N. 1976. Breast patterns as an index of risk for developing breast cancer. American Journal of Roentgenology, 126, 1130-1139. WOLFE, J. N., ALBERT, S., BELLE, S. & BALANE, M., 1983.
Breast patterns and their relationship to risk for having or developing carcinoma. Radiological Clinics of North America, 21, 127-136. WOLFE,
J. N.,
SAFTLAS,
A. F.
&
SALANE,
M.,
1987.
Mammographic parenchymal patterns and quantitative evaluation of mammographic densities: a case-control study. American Journal of Roentgenology, 148, 1087-1092.
pathology of the human breast with special reference to
Vol. 63, No. 755
849
The relationship of "high risk" mammographic patterns to histological risk factors for development of cancer in the human breast By Justine E. Arthur, MB, BS, I. O. Ellis, MB, BS, MRCPath, *C. Flowers, MRCS, LRCP, FRCR, *E. Roebuck, DMRD, FRCR, C. W. Elston, MD, MRCPath and tR. W. Blarney, MD, FRCS Departments of Histopathology and tSurgery, City Hospital and *Department of Radiology, University Hospital, Nottingham {Received February 1990 and in revised form May 1990)
Abstract. In the UK Trial for the Early Detection of Breast Cancer in Nottingham, 119 women were identified as having fibrocystic change with epithelial hyperplasia or in situ carcinoma. Their mammograms were classified according to Wolfe's criteria and the corresponding histology for each patient was classified for degrees of epithelial hyperplasia, atypia and in situ neoplasia using Page's criteria. A control population of patients presenting for breast screening was used to represent the general population. No correlation was found between the four mammographic Wolfe patterns, N l , PI, P2 and DY and histological evidence of epithelial hyperplasia, atypia or in situ carcinoma. A further study was carried out to determine histological features of Wolfe pattern, using radiological examination of resected breast tissue. The variation in Wolfe pattern was related to the distribution of fibrous and adipose tissue in the breast interlobular stroma and appeared to have no relationship to epithelial parenchymal content. This information does not support the hypothesis that radiographic densities of P2 and DY patterns correspond to high risk epithelial proliferation.
Each year 15 000 women in England and Wales die from breast cancer and 1 in 15 women will develop the disease in their lifetime (OPCS, 1983, HMSO, 1983). The current lack of any prospect for primary prevention increases the need to develop more effective and less expensive techniques for early detection of breast cancer. Selection of women at high risk might reduce the number who need to attend for regular mammographic screening. The work of Wolfe shows that women exhibiting certain mammographic breast parenchymal patterns are at greater risk of developing breast cancer (Wolfe, 1976, 1987; Wolfe et al, 1983). He has described the following breast parenchymal patterns: Nl the breast is composed primarily of fat, often with a fine trabeculated appearance; PI "prominent" ducts occupy 25% or less of the volume of the breast; P2 "prominent" ducts occupy more than 25% of the volume of the breast; DY mammary dysplasia (fibrocystic change) with sheet-like areas of increased density. Wolfe states that the mammographic patterns most often related to breast cancer and premalignant epithelial alterations, especially after the age of 50, are the P2 and DY breast patterns, Nl and PI being rarely associated with these conditions. He reports that the incidence of subsequent carcinoma development, per 1000 women, over a period of 3 years increases 30 fold, from Address for correspondence: Dr I. O. Ellis, Department of Histolopathology, City Hospital, Hucknall Road, Nottingham NG5 1PB. Vol. 63, No. 755
1.45 cases per 1000 women with Nl patterns to 44.60 cases per 1000 women with DY patterns. (PI = 3.85 and P2 = 17.45) Several categories of epithelial proliferation have been described in the female breast (Page & Anderson, 1988): regular epithelial hyperplasia of mild, moderate and florid degree, atypical ductal hyperplasia and atypical lobular hyperplasia; and ductal and lobular carcinoma in situ. Increasing risks of subsequent development of breast carcinoma have been demonstrated with these conditions (Page et al, 1978, 1982, 1985). Mild hyperplasia carries the same risk as the general population of the subsequent development of invasive breast carcinoma. Moderate and florid hyperplasia carry a risk double that of the general population for cancer development. Atypical hyperplasia of ductal and lobular types are associated with four times the risk for subsequent development of invasive carcinoma. Ductal and lobular carcinomas in situ are associated with a risk for invasive carcinoma development 10 times that of the general population (Hutter et al, 1986; Page et al, 1978, 1982; Page & Anderson, 1988). Despite these detailed studies of histopathological and radiological risks for carcinoma development in the breast, little work has been undertaken to relate these phenomena to each other. One study, performed by Wellings and Wolfe (1978) correlated the radiographic densities of PI, P2 and DY mammographic patterns withfibrosisand a variety of lesions of the terminal duct lobular units particularly "atypical lobules Type A". Two of these lesions were described previously by Wellings and Jensen (1975) but both were thought to arise from a similar histogenetic pathway. Type A 845
Justine E. Arthur et al Table I. The study population identified from the Nottingham Breast Self Examination (BSE) arm of the UK Trial for the Early Detection of Breast Cancer 1979-1986
Method
Between 1979 and 1986, 56000 women aged 45-64 years in Nottingham were invited for education in Breast Self Examination as part of the United Kingdom Women attending for Trial of Early Detection of Breast Cancer. Women BSE 28 759 (49.86% of those invited) participating in this programme were offered an open Number of biopsies in access mammography and clinical examintion service the study group for should they detect a breast abnormality through self malignant conditions 546 ~) examination. If a significant lesion was identified at Number of biopsies in >119 borderline biopsies* the study group for clinic attendance, open biopsy was performed. During benign conditions 449J this study period biopsies were carried out on 995 of the study group. One hundred and nineteen cases were * Borderline biopsies indicates the number of biopsies for identified as "borderline lesions" of epithelial hypercarcinoma in situ (included in the malignant category) and plasia and in situ carcinoma (Table I). epithelial hyperplasia (included in the benign category). In our study, these 119 patients with borderline lesions were taken to represent women with histological evidence of increased risk. Forty-three women who lesions represent the spectrum of ductal hyperplasia, underwent biopsies which showed benign non-proliferaatypical ductal hyperplasia and ductal carcinoma in situ. tive fibrocystic change are taken to represent women Type B lesions represent the lobular equivalents, that is, with no histological evidence of increased risk atypical lobular hyperplasia and lobular carcinoma (Table II). To represent the general population, 91 in situ. Wellings and Wolfe (1978) found that the degree women, aged between 50 and 64, consecutively of fibrosis and the grade of Type A atypical lobules attending for mammographic breast screening at the increased from PI to DY. A recent study by Urbanski et Nottingham City Hospital during 1998, were used to al (1988) demonstrated a weak relationship between the determine the frequency of each mammographic extent of mammographic "dysplasia" and histological pattern. All mammograms on the patients were availevidence of epithelial atypia and carcinoma in situ in able. The Wolfe classification of all these mammograms women younger than 50 years old. Such findings could was reviewed by a single radiologist to elimiate interexplain the increased risk for cancer development asso- observer variation in the objective grading of mammociated with P2 and DY breasts. grams. The radiological classification based on the We have investigated this putative relationship mammographic pattern was then compared with the between Wolfe pattern and epithelial proliferation with histological findings. Improvement of mammographic two studies. First, a retrospective study was carried out technique has occurred between 1979 and 1988. of the mammographic appearances and histology of However, the differences was not considered sufficient to women from the Nottingham arm of the UK Trial for alter interpretation of the Wolfe pattern. the Early Detection of Breast Cancer. Secondly a The histological slides were stained in Ehrlich's prospective study was performed combining radiolo- haematoxylin and eosin (H&E), and were assessed by gical and histological examination of mastectomy speci- two pathologists using Page's method of classification mens from women undergoing treatment for breast (Page & Anderson, 1988). Mild regular hyperplasia is carcinoma. defined by the recognition of between two and four
Table II. Distribution of the cases according to Wolfe mammographic pattern categories (percentages in parentheses) Screened population
FCC
Nl PI P2 DY
15 12 16 48
(16) (13) (18) (53)
1 (2) 9 (21) 15 (35) 18 (42)
Total
91 (100)
43 (100)
All pathological "borderline" groups
Pathological "borderline" groups Mild EH
Mod EH
(11) (20) (27) (42)
0 (0) 2 (12) 3 (19) 11 (69)
3 (10) 8 (28) 9 (31) 9 (31)
119 (100)
16 (100)
29 (100)
13 24 32 50
ADH 0 0 2 3
(0) (0) (40) (60)
5 (100)
DCIS 8 12 15 23
(14) (21) (26) (39)
58 (100)
ALH 0 (0) 0 (0) 0 (0) 0 (0) 0
(0)
LCIS 2 2 3 4
(18) (18) (27) (36)
11 (100)
FCC = fibrocystic change. Mild EH = mild epithelial hyperplasia. Mod EH = moderate and florid epithelial hyperplasia. ADH = atypical ductal hyperplasia. DCIS = ductal carcinoma in situ. ALH = atypical lobular hyperplasia. LCIS = lobular carcinoma in situ.
846
The British Journal of Radiology, November 1990
"High risk" mammographic patterns and histological risk factors in breast cancer ] _„
Mammographic Parenchymal •
80
Pattern
60-
50-
40-
30-
10-
30-
II A
B
CD
E
F
N1 P1 P2 Mammographic Parenchymal Pattern
Histological Risk Group
Figure 1. The distribution of Wolfe mammographic breast parenchymal patterns in various histological subgroups. A: Screened/general population; B: non-proliferative fibrocystic change; C: mild regular epithelial hyperplasia; D: moderate florid regular epithelial hyperplasia, mild atypical hyperplasia; E: atypical epithelial hyperplasia; F: carcinoma in situ (ductal and lobular).
epithelial cells above the basement membrane. Moderate and florid regular hyperplasia are recognized as five or more cells above the basement membrane, with a tendency to distension and crossing of the space by hyperplastic cells which do not show the cytological and architectural patterns of atypical hyperplasia and ductal carcinoma in situ. Atypical ductal hyperplasia is recognized by its histological pattern which is intermediate between that of regular hyperplasia and ductal carcinoma in situ, that is, with some but not all the histological features of ductal carcinoma in situ. Ductal carcinoma in situ in its various subtypes is recognized by its characteristic architectural and cytological features, well described by Page and Anderson (1988). Atypical lobular hyperplasia and lobular carcinoma in situ are recognized by an expansion of small uniform cells with lobular units which may extend to involve ducts in a Pagetoid manner. Filling of all acini within the lobule with distortion of at least 50% of the lobule is used as the arbitrary divide between the two conditions. An additional prospective study was also performed involving a radiographic and histological study of breast slices from 17 consecutive women who had undergone mastectomy. All of these patients had had pre-operative mammography. It is not possible to examine the entire breast in a histologicaJ section. Specimen radiography was used to confirm that the areas of breast sampled had a radiographic appearance representative of the mammographic pattern. Slices measuring 1-2 cm in Vol. 63, No. 755
Figure 2. The distribution of the normal population and histological risk groups found in each of the Wolfe mammographic parenchymal patterns. A-E: as in Fig. 1; F: ductal carcinoma in situ; G: lobular carcinoma in situ.
thickness were taken from each mastectomy specimen. The slice included the nipple and was taken from as near as possible to the 3-9 o'clock position in the horizontal plane. A specimen radiograph was taken and correlated with the pre-operative cranio-caudal mammographic view. Using the specimen radiograph and mammogram it was possible to sample areas of tissue containing the radiographic densities characteristic of the mammographic pattern. On average, 10 blocks were taken from each slice, and between four and six sections were cut and stained in Ehrlich's H & E. The direction and position of each block was recorded on diagrams corresponding with the mammograms and specimen radiographs. Results
The distribution of breast patterns seen in the control group and in each histological group is shown in Table II and illustrated in Fig. 1. There was no correlation between the radiological and the histological risk factors for development of breast cancer. Figure 2 shows the overall distribution of the histological conditions seen in each mammographic subgroup. These results demonstrate that any increased risk associated to P2 and DY patterns cannot be attributed to a greater proportion of high risk epithelial abnormalities in the breasts. From the study of specimen radiology of women who had undergone mastectomy, it appears that the variation in Wolfe pattern was a result of the ratio of fibrous to adipose tissue in the breast with no apparent relationship to epithelial parenchymal content. A varying distribution of collagenous fibrous tissue and adipose tissue 847
Justine E. Arthur et al
was observed in the breasts examined. A classification of this distribution based on the arrangement of these tissues was devised. Three basic patterns were observed: (1) septae, where fibrous tissue was confined to thin fibrous septae usually containing vessels and subsegmental ducts; (2) islands, where the fibrous tissue was arranged in nodular collections, usually containing parenchymal structures, and surrounded by adipose tissue; (3) confluent, where the fibrous tissue was arranged as a diffuse sheet with little adipose tissue present. Within some breasts, more than one type of distribution was observed across the breast disc. The dominant pattern was recorded in these instances. The results of this aspect of the study are shown in Table III. In addition, the proportion of epithelial parenchymal structures present was recorded. This was highly variable between cases with no apparent relationship to mammographic background pattern or histological fibrous tissue and adipose tissue distribution. Discussion We have examined the putative link between mammographic breast parenchymal pattern and histological evidence of epithelial proliferation, a documented risk factor for the development of breast cancer. If the high risk associated with P2 and DY Wolfe patterns is to be attributed to high grades of epithelial premalignant abnormality, one would expect the proportion of these high risk mammographic patterns to be increased in the high risk histological groups. No such distribution is seen in this study and our findings do not support the results reported by Wellings and Wolfe (1978). In fact, the ratio of high risk (P2 and DY) to low risk (Nl and PI) mammographic patterns is increased in the low risk histology control group and screening control group compared with in the high risk histological groups of in situ carcinoma. In all the histological groups, and in the two control populations, the commonest mammographic categories were P2 and DY (Figs 1 and 2). This observation, although well recognized by radiologists, is not well documented. It could provide an alternative explanation why an increased incidence of carcinoma was demonstrated with these two mammographic patterns. The failure of this study to confirm a correlation between Wolfe mammographic parenchymal pattern and histological epithelial alteration suggests that any high risk for carcinoma development associated with P2 and DY patterns is not due to an epithelial abnormality. Instead, the variation in mammographic patterns appears to be related to the variable ratio and distribution of fibrous tissue and adipose tissue in the breast interlobular stroma. Nl patterns reflect a stroma predominantly composed of adipose tissue with only thin fibrous tissue septae being present. The PI and P2 patterns result from stroma composed of islands and septae of fibrous tissue within adipose tissue. In DY patterns there is coalescence of extensive fibrous islands and septae to produce large areas of virtual confluence of fibrous connective tissue (Table III). 848
Table III. Histological distribution of fibrous tissue within Wolfe radiological groups Breast Pattern
Septae
Islands*
Confluent*
Nl (n = 1) PI (« = 1) P2 (n = 8) DY (« = 7)
1 0 4 3
0 1 8 4
0 0 2 7
* Variable epithelial parenchymal content.
These descriptions of the association of high risk mammographic patterns with the distribution of fibrous tissue are consistent with the findings of Fisher et al (1978) and Mansel (personal communication), but are contrary to prevailing views that associate the radiographic densities of P2 and DY patterns and their corresponding risk for breast cancer development with proliferative fibrocystic change (Wellings & Wolfe, 1978). Although Urbanski et al (1988) have demonstrated a significant association between epithelial atypia and carcinoma in situ and mammographic "dysplasia", this associaton was very weak and could not be reproduced on independent review by a second radiologist. They recognized that the epithelial changes observed were not directly responsible for the mammographic findings. We are unable in this study to address directly the association between subsequent risk of breast cancer and mammographic background pattern. Our observation that these radiographic densities, which are characteristic of the four mammographic patterns, are simply a result of variation in fibrous tissue in the interlobular stroma does not necessarily dispel claims of risk associated with P2 and DY patterns. To our knowledge there are no studies which have investigated the distribution of stromal components to risk or development of breast cancer. Further evaluation of the radiographic determinants of risk and these stromal alterations in the development of breast carcinoma is required. Acknowledgments We wish to thank the administrative staff of the Nottingham Breast Screening Service for their help in identification of the cases studied, the audiovisual department for assistance in preparing the figures and Julie Rainbow and Jinny Spence for preparing the manuscript. References FISHER, E. R, PALEKAR, A., KIM, W. S. & REDMOND, C ,
1978.
The histopathology of mammographic patterns. American Journal of Clinical Pathology, 69, 241-426. HMSO, 1983. Cancer Statistic Registrations (HMSO, London). HUTTER, R. P. V., 1986. Consensus meeting: Is 'fibrocystic disease' of the breast precancerous? Archives of Pathology and Laboratory Medicine, 110, 171-173. OPCS, 1983. Cancer Statistics Registrations (OPCS, England and Wales). PAGE,
D. L.
&
Histopathology Edinburgh).
ANDERSON,
of
the
Breast
T. J.,
1988.
(Churchill
Diagnostic
Livingstone,
The British Journal of Radiology, November 1990
"High risk" mammographic patterns and histological risk factors in breast cancer PAGE, D. L., ZWAGG, R. V., ROGERS, L. W., WILLIAMS, L. T., WALKER, W. E. & HARTMANN, W. H., 1978. Relation
precancerous lesions. Journal of the National Cancer Institute, 55, 231-273.
between the component parts of nbrocystic disease complex and breast cancer. Journal of the National Cancer Institute, 61, 1055-1063.
WELLINGS, S. R. & WOLFE, J. N., 1978. Correlative studies of
PAGE,
D. L.,
DUPONT,
W. D.,
ROGERS,
L. W.
&
LANDENBERGER, M., 1982. Intraductal carcinoma of the breast: follow-up after biopsy. Cancer, 49, 751-758. PAGE, D. L., DUPONT, W. D., ROGERS, L. W., WILLIAMS, L. T., WALKER, W. E. & HARTMANN, W. H., 1985. Atypical
hyperplastic lesions of the female breast—long term follow study. Cancer, 55, 2698-2708. URBANSKI, S., JENSEN, H. M., COOKE, G., MCFARLANE, D., SHANNON, P., KRUIKOV, V. & BOYD, N. F., 1988. The
association of histological and radiological indicators of breast cancer risk. British Journal of Cancer, 58, 474-479. WELLINGS, S. R. & JENSEN, H. M., 1979. An atlas of subgross
the histological and radiographic appearances of the breast with special reference to possible precancerous lesions. Diagnostic Radiology, 129, 299-306. WOLFE, J. N. 1976. Breast patterns as an index of risk for developing breast cancer. American Journal of Roentgenology, 126, 1130-1139. WOLFE, J. N., ALBERT, S., BELLE, S. & BALANE, M., 1983.
Breast patterns and their relationship to risk for having or developing carcinoma. Radiological Clinics of North America, 21, 127-136. WOLFE,
J. N.,
SAFTLAS,
A. F.
&
SALANE,
M.,
1987.
Mammographic parenchymal patterns and quantitative evaluation of mammographic densities: a case-control study. American Journal of Roentgenology, 148, 1087-1092.
pathology of the human breast with special reference to
Vol. 63, No. 755
849
