Professional Documents
Culture Documents
SUMMARY
Recent experimental evidence obtained in Scid mice has suggested that the metastatic process is in large part epigenetically regulated
and undergoes partial reversion once the metastatic process is completed: the metastatic colonies become more engaged in the process
of growing in situ than actively metastasizing. Based on this experimental evidence, examples were sought of metastatic human cancers
where similar reversion to an in situ growth state was occurring. Review of 200 cases of metastatic human breast cancer revealed a 21
per cent incidence of reversion to a ductal carcinoma in situ (DCIS) growth pattern within axillary nodal metastases. The revertant
DCIS areas were characterized by an intact and circumferential basement membrane, as demonstrated by extracellular laminin and type
IV collagen immunoreactivity. These revertant DCIS areas could be distinguished from primary DCIS, however, by the absence of
surrounding myoepithelial cells in the former, identified in the latter by their positive maspin, S-100, and smooth muscle actin
immunoreactivity. The pattern of revertant DCIS, poorly differentiated (comedo) (13 per cent), intermediate (non-comedo) (6 per cent),
or well-differentiated (non-comedo) (2%), exhibited complete 100 per cent concordance with the primary DCIS pattern. The
concordance of histological patterns held true for even the subtypes of DCIS determined by architectural pattern, such as the
micropapillary or cribriform subtypes. Nuclear size by digital image analysis and Her-2/neu, p53, and Ki-67 status in the revertant DCIS
also exhibited complete concordance with the primary DCIS counterparts. Cases exhibiting a revertant DCIS pattern tended to be
ER-negative/EGFR-positive and exhibited significant nodal involvement (mean number, 9; mean area, 90 per cent) compared with cases
lacking a revertant pattern (mean number, 4; mean area, 15 per cent) (P<0·01) These findings suggest that reversion of the metastatic
phenotype may also be occurring within autochthonous human metastasis. ? 1997 John Wiley & Sons, Ltd.
Diagnostic criteria—Modern diagnostic criteria for brownish-black staining) at sites of antigen presence.
the diagnosis and classification of DCIS types and Nuclei of cells containing ER protein, Ki-67, and p53
subtypes as established by recent studies4,5 were stained; surface membranes of cells containing either
applied. Specifically, DCIS was classified into poorly increased Her-2/neu antigen or increased EGFR antigen
differentiated, intermediately differentiated, and well- stained. For ER, tissues with greater than 10 per cent
differentiated on the basis of cytonuclear differentiation. nuclear staining (weak to strong staining intensity) were
The presence or absence of central necrosis was not used considered positive. For p53, EGFR, and Her-2/neu
as a basis of this classification but the term comedo was staining, either the cells were uniformly negative or
applied to poorly differentiated cases and the term 25–100 per cent staining occurred, the latter pattern of
non-comedo was applied to both intermediately and which was considered positive. For Ki-67, the number of
well-differentiated cases. DCIS was also subclassified positive nuclei was expressed as a percentage of the total
on the basis of architectural differentiation into solid, nuclei. For all of these antigens, cytoplasmic granular
cribriform, and micropapillary subtypes. staining was considered non-specific.
Not only did the revertant DCIS growth pattern genetic locus at chromosome 11q13 is altered in both
closely recapitulate the primary DCIS pattern by histo- microdissected in situ and invasive breast cancer from
pathological criteria, but by digital image analysis, the the same patient.20 In retrospective studies, pathologists
nuclear sizes exhibited by the two DCIS areas were have noted an approximately 25 per cent incidence of
essentially identical (P>0·25) (Fig. 2A). Strong concord- progression of DCIS to invasive cancer over a 6 to
ance of Ki-67, Her-2/neu, and p53 status between the 10-year period,21,22 but whether additional genetic
revertant DCIS and the primary DCIS within individual events in the DCIS cells or paracrine events governed
cases was also observed. When the individual patients by the myoepithelial or other host cells influence this
were grouped into primary, revertant, poorly differenti- progression is not known.23
ated (comedo), intermediately, and well-differentiated The cardinal histopathological distinction between
(non-comedo), no differences were observed in these DCIS of all types and invasive breast cancer is the
markers between the revertant and primary DCIS presence of a circumferential or near-circumferential
(P>0·25), but significant differences were observed basement membrane and myoepithelial cell layer around
between the comedo and non-comedo groups (P<0·05) the ducts involved by DCIS.7,24 The myoepithelial cells
(Figs 2B and 2C). ER and EGFR status in the revertant are thought to contribute largely to the synthesis of the
DCIS exhibited strong concordance with their primary basement membrane and to exert other important para-
DCIS pattern (Fig. 2C). The groups exhibiting revertant crine influences on both the overlying normal ductal
primary comedo DCIS showed a higher Ki-67 status epithelium and the DCIS. Even when the DCIS involves
(Fig. 2B) and a greater presence of p53 and Her-2/neu the adjacent lobules (so-called ‘cancerization’ of the
alterations (Fig. 2C) compared with the groups exhibit- lobules), the surrounding basement membrane and
ing revertant and primary non-comedo DCIS (P<0·05). myoepithelial layer largely remain intact. In invasive
Cases exhibiting revertant DCIS were largely breast cancer, on the other hand, the invading breast
ER-negative and EGFR-positive (Fig. 2C), irrespective cancer cells penetrate this basement membrane and
of whether their pattern was revertant comedo or myoepithelial layer. Although some carcinomas, such as
non-comedo DCIS. squamous cell carcinomas, can retain a surrounding
Cases exhibiting a revertant DCIS pattern exhibited basement membrane during active invasion,25 invasive
significant nodal involvement (mean number, 9; mean breast carcinomas of all types, including even the most
area, 90 per cent), irrespective of whether the revertant differentiated tubular carcinomas, consistently lack a
DCIS pattern was comedo or non-comedo, compared surrounding basement membrane. The observation in
with cases not exhibiting revertant DCIS (mean number, this study of a recapitulation of basement membrane
4; mean area 15 per cent) (P<0·01) (Fig. 2D). In these around the islands of metastatic breast carcinoma cells,
cases of revertant DCIS, the majority of the node was cells which lack surrounding myoepithelial cells, is com-
replaced by metastatic carcinoma and the revertant pelling evidence that the carcinoma cells have acquired
DCIS pattern was observed in 15–25 per cent of the an ‘in situ’ phenotype. The strong correlation of his-
node. In many cases of revertant DCIS, the overall topathological features, digital image analysis nuclear
appearance of the nodal metastases bore a strong resem- parameters, and molecular markers between the rever-
blance to that of the primary tumour, which usually tant and primary DCIS areas of individual cases indi-
showed extensive intraductal carcinoma (EIC) admixed cates that this phenomenon is not stochastic but highly
with invasive carcinoma. In all of the 200 cases, most of directed.
which had readily apparent primary DCIS or EIC, in These findings have value in our understanding of the
the primary tumour either a continuous or a focally metastatic process in human breast cancer, because the
discontinuous layer of myoepithelial cells was always recognition of the revertant DCIS phenomenon sup-
present around the islands of tumour cells showing ports the hypothesis that metastatic carcinomas can
extracellular basement membrane staining. We never recreate their primary microenvironment at a distant site
observed extracellular basement membrane staining and engage in in situ growth rather than subsequent
around any tumour island exhibiting a complete absence metastasis. This hypothesis receives additional support
of myoepithelial cells in the primary tumour. from many other observations that have been made in
diverse tumours, such as the presence of sustentacular
cells at the periphery of the ‘Zellballen’ of invasive and
DISCUSSION metastatic paraganglioma, the occurrence of follicular
dendritic cells in the nodules of follicular lymphoma that
Breast cancer development in vivo includes not only a has spread outside the lymph node, and the presence of
derangement of growth, but also a progression to an non-neoplastic T-lymphocytes in between the tumour
invasive and highly metastatic phenotype.18,19 The cells of metastatic thymoma. In the case of metastatic
molecular and biochemical mechanisms behind this breast carcinoma, the attempt to recreate the micro-
progression are not well defined, but for decades path- environment of the primary tumour at a distant site
ologists have felt that ductal carcinoma in situ (DCIS), a manifests itself as a reversion to a DCIS growth state.
preinvasive proliferative growth state present within the Interestingly, the phenomenon that we are describing
primary ductal lobular units of the breast, progresses to is well recognized anecdotally among pathologists.
invasive and subsequently metastatic breast cancer. When we showed our cases to our pathology colleagues
Recently, this belief has been given credence by loss of at our institution, many responded that they had
heterozygosity (LOH) studies observing that the same ‘seen these patterns before’. We recently learned that
? 1997 John Wiley & Sons, Ltd. , . 183: 188–194 (1997)
REVERTANT DCIS IN AUTOCHTHONOUS METASTASIS 191
Fig. 1—A–F
Fig. 1—(A) Revertant DCIS pattern (left) is seen juxtaposed to invasive areas (right) in lymph node metastases. Revertant DCIS patterns
included (B) poorly differentiated (comedo); (C) well-differentiated solid, (D) well-differentiated cribriform; (E) intermediately differentiated
micropapillary. There was strict fidelity between the type and subtype of primary DCIS and revertant DCIS. Some cases of nodal revertant
DCIS showed dystrophic luminal calcifications (F). All of the patterns of nodel revertant DCIS were surrounded by an intact basement
membrane, as evidenced by strong immunoreactivity to both type IV collagen and laminin (depicted) (G). Adjacent non-DCIS areas were
completely devoid of this extracellular immunoreactivity. The patterns of nodal revertant DCIS were, however, devoid of surrounding
myoepithelial cells (H), which were consistently present around primary DCIS of all types (I). Myoepithelial cells could be identified by
maspin (depicted), S-100 or smooth muscle actin positivity. (A, B, C, D, E, F) H & E, #150–#400; (G, H, I) Anti-laminin/anti-maspin,
immunoperoxidase, #250–#400
Professor N. F. Gowing, while working at the Royal observed that our experimental metastases did not
Marsden Hospital in the 1970s, made similar observa- engage in subsequent metastasis, but rather engaged in
tions of a DCIS growth pattern in axillary lymph nodes growing in situ. Based on these experimental studies, we
using PAS staining for basement membranes (personal reasoned that a large component of the metastatic
communication). Recently we published an experimental process was epigenetically regulated and hence revers-
study1 where we examined the issues of the paracrine ible. Based on this earlier experimental study which we
regulation of the metastatic process. In that study we conducted in Scid mice, we began looking for examples
? 1997 John Wiley & Sons, Ltd. , . 183: 188–194 (1997)
192 S. H. BARSKY ET AL.
Fig. 1—G–I
in human pathological material that supported our cent certainty as being a true carcinoma ‘in situ’.26
overall hypothesis. The present study is an observational We did not observe in any of our 200 cases examples
account of a DCIS pattern of growth in metastases of of islands of invasive breast carcinoma lacking myo-
human breast cancer which suggests that reversion to a epithelial cells but exhibiting basement membrane
non-invasive in situ growth state may also be occurring immunoreactivity in the primary tumour.
in human autochthonous metastasis. We cannot be Another potential concern was whether the revertant
sure, however, that this growth pattern represents true DCIS pattern that we were observing actually repre-
biological reversion. sented tumour cells lodged in basement membrane-
Despite this, we believe that the areas of ‘revertant’ containing lymph node sinusoids. We considered this
DCIS represent more than duct-like differentiation in an alternative explanation for our observations, but
invasive tumour. The proof of this argument lies in the decided that it was unlikely for the following reasons:
consistent presence of surrounding basement mem- firstly, we did not see membrane-bound structures
branes around these islands of revertant DCIS. Even the present in metastatic deposits outside nodes. Secondly,
most differentiated invasive ductal carcinomas, such as in virtually all of the cases, the areas of revertant DCIS
tubular carcinomas, lack surrounding basement mem- were present in medullary and cortical-medullary areas
brane structures. Hence duct-like differentiation per se is of the lymph node, sparing the subcortical lymphatic
not sufficient to recapitulate basement membrane immu- spaces which were visible but not involved by metastatic
noreactivity. Furthermore, in the cases demonstrating tumour. We felt it unlikely for metastatic deposits within
nodal revertant DCIS, analysis of the primary lesions nodal lymphatics to involve selectively only medullary
revealed no evidence of similar structures containing nodal areas, sparing subcortical lymphatics. Thirdly, in
basement membranes but lacking myoepithelial cells. a substantial fraction of cases the nodes contained a
We are aware that some cases of primary breast cancer desmoplastic reaction and the pattern of revertant DCIS
might exhibit areas of a DCIS-like pattern which are was contained within these desmoplastic areas. When
nevertheless invasive and lack surrounding myoepithe- metastases occupy nodal lymphatics or when there is a
lial cells and basement membrane immunoreactivity. sinus histiocytosis involving nodal lymphatics, there
This could explain the finite risk of nodal metastasis in usually is not a desmoplastic reaction.
apparently ‘pure’ DCIS. This possibility was first sug- Our studies indicated that many of the known
gested in the 1960s by Professor Carlo Sirtori, who markers of breast carcinoma progression, Her-2/neu,
argued in the Italian literature that intraductal carci- p53, loss of ER, increased EGFR expression, and
noma of the breast could never be regarded with 100 per increased Ki-67, exhibited a strong correlation between
? 1997 John Wiley & Sons, Ltd. , . 183: 188–194 (1997)
REVERTANT DCIS IN AUTOCHTHONOUS METASTASIS 193
Fig. 2—(A) Nuclear size (mean&standard deviation) strongly correlated between the revertant and primary DCIS areas; poorly differentiated DCIS
(comedo) exhibited a larger nuclear size than intermediately and well-differentiated DCIS (non-comedo). (B ) Correlation of Ki-67 immunoreactivity
(mean percentage of positivity&standard deviation) was strong between revertant and primary DCIS for both poorly differentiated (comedo) and
intermediately and well-differentiated (non-comedo) DCIS, which, differed significantly from each other. (C) Correlation between revertant DCIS
(R-DCIS) and primary DCIS (P-DCIS) in the expression (percentage of cases where marker was positive) of different molecular markers implicated
in human breast carcinoma progression. (D) Nodal involvement: the number of nodes and the percentage of nodal area (mean&standard deviation)
were greater in cases in which a pattern of revertant DCIS was observed, suggesting that a metastatic size threshold is required to initiate the
revertant phenotype
the revertant and primary DCIS areas and hence altera- not usually observed when the metastatic tumour
tions in these markers were not the cause of the DCIS volume/mass falls below a certain threshold. Cases
revertant phenotype. As anticipated, both revertant and which are EGFR-positive/ER-negative represent a sub-
primary poorly differentiated (comedo) DCIS exhibited set of breast carcinoma cases which exhibit more
an increase in nuclear size, the Ki-67 proliferating aggressive biological behaviour and more lymph node
marker, and frequency of p53 and Her-2/neu alterations metastases than cases which are EGFR-negative/
compared with their intermediately and well- ER-positive.
differentiated (non-comedo) DCIS counterparts.27,28 Since no internal cellular (nuclear, cytoplasmic, or
Cases of revertant DCIS tended to be EGFR--positive membrane) markers exist for breast carcinoma which
and ER-negative, irrespective of whether their pattern reliably distinguish DCIS cells from invasive carcinoma
was comedo or non-comedo; furthermore, cases of cells, we could only rely upon the extracellular basement
revertant DCIS, irrespective of whether their pattern membrane markers, laminin and type IV collagen, to
was comedo or non-comedo, exhibited significant nodal distinguish invasive from DCIS (primary or revertant)
involvement in terms of both the number of nodes and phenotypes. Recent in situ hybridization studies of
the percentage of nodal area involved. These observa- the calcium-binding protein psoriasin have observed
tions indicated to us that the revertant phenomenon is a increased expression in the in situ versus the in-
function of the metastatic tumour volume/mass and is vasive component of the same breast tumour.29 In
? 1997 John Wiley & Sons, Ltd. , . 183: 188–194 (1997)
194 S. H. BARSKY ET AL.
collaborative studies we are investigating whether 7. Barsky SH, Siegal GP, Jannotta F, Liotta LA. Loss of basement membrane
components by invasive tumors but not by their benign counterparts. Lab
increased psoriasin expression occurs in revertant DCIS. Invest 1983; 49: 140–148.
Such a demonstration would provide additional proof of 8. Cossman J, Schlegal R. p53 in the diagnosis of human neoplasia. J Natl
a true reversion to a DCIS state. Cancer Inst 1991; 18: 980–981.
9. Visakorpi T, Kallioniemi OP, Koivula T, Harvey J, Isola J. Expression of
Previous detailed pathological studies on the mor- epidermal growth factor receptor and ERB2 (Her-2/neu) oncoprotein in
phology of axillary metastases of human breast cancer prostatic carcinomas. Mod Pathol 1992; 5: 643–648.
10. Cattoretti G, Becker MH, Key G, et al. Monoclonal antibodies against
have indicated strong morphological identity in terms of recombinant parts of the Ki67 antigen (MIB 1 and MIB 3) detect prolifer-
glandular differentiation between the primary invasive ating cells in microwave-processed formalin-fixed paraffin sections. J Pathol
carcinoma and the lymph node metastases.30 These 1992; 168: 357–363.
11. Allred D, Clark G, Molina R, et al. Overexpression of Her-2/neu and its
studies have indicated that during human breast carci- relationship with other prognostic factors change during the progression of
noma metastasis, selection for morphologically more in situ to invasive breast cancer. Hum Pathol 1992; 23: 974–979.
aggressive clones does not occur. Our present observa- 12. Cheng L, Binder S, Fu Y, Lewin K. Demonstration of estrogen receptors by
monoclonal antibody in formalin-fixed breast tumors. Lab Invest 1988; 58:
tions extend these findings. Not only do axillary meta- 346–353.
stases exhibit morphological identity with their primary 13. Greene G, Nolan C, Engler J, Jensen E. Monoclonal antibodies to human
estrogen receptor. Proc Natl Acad Sci USA 1990; 77: 5115–5119.
invasive carcinoma counterparts, but high volume/mass 14. Shana-Rong S, Key ME, Kalra KL. Antigen retrieval in formalin-fixed
axillary metastases exhibit morphological identity with paraffin-embedded tissues: an enhancement method of immunohistochemi-
their primary DCIS counterparts as well. This suggests cal staining based on microwave oven heating of tissue sections. J Histo-
chem Cytochem 1991; 39: 741–748.
that axillary metastases are not the product of clonal or 15. Barsky SH, Layfield L, Varki N, Bhuta S. Two human tumors with high
genetic selection, but rather represent cellular popula- basement membrane-producing potential. Cancer 1988; 61: 1798–1806.
tions epigenetically derived from the primary tumour, 16. Sternlicht MD, Safarians S, Rivera SP, Barsky SH. Characterizations of the
extracellular matrix and proteinase inhibitor content of human myoepi-
capable of reverting back to their in situ growth state. thelial tumors. Lab Invest 1996; 74: 781–796.
17. Zou Z, Anisowicz A, Hendrix MJC, et al. Maspin, a serpin with tumor-
suppressing activity in human mammary epithelial cells. Science 1994; 263:
ACKNOWLEDGEMENTS 526–529.
18. Kohn EC, Liotta LA. Molecular insights into cancer invasion: strategies for
prevention and intervention. Cancer Res 1995; 55: 1856–1862.
This work was supported by USPHS grants CA71195, 19. Liotta L, Steeg P, Stetler-Stevenson W. Cancer metastasis and angiogenesis:
CA40225, and CA01351, and funds from the California an imbalance of positive and negative regulation. Cell 1991; 64: 327–336.
Institute for Cancer Research (CICR), the Cancer 20. Zhuang Z, Merino MJ, Chuaqui R, Liotta LA, Emmert-Buck MR. Identi-
cal allelic loss on chromosome 11q13 in microdissected in situ and invasive
Research Coordinating Committee (CRCC), and the human breast cancer. Cancer Res 1995; 55: 467–471.
Jonsson Comprehensive Cancer Center (JCCC). Dr 21. Page D, Dupont W, Rogers L, Landenberger M. Intraductal carcinoma of
Doberneck is a recipient of an American Cancer Society the breast: follow-up after biopsy only. Cancer 1982; 49: 751–758.
22. Rosen P, Braun D, Kinne D. The clinical significance of pre-invasive breast
Fellowship. Dr Barsky is a recipient of a NCI Research carcinoma. Cancer 1980; 46: 919–925.
Career Development Award. 23. Sternlicht MD, Safarians S, Calcaterra TC, Barsky SH. Establishment and
characterization of a novel human myoepithelial cell line and matrix
producing xenograft from a parotid basal cell adenocarcinoma. In Vitro Cell
Dev Biol 1996; 32: 550–563.
REFERENCES 24. Lagios MD, Westdahl PR,. Margolin FR, Rose MR. Duct carcinoma in
situ. Cancer 1982; 50: 1309–1314.
1. Safarians S, Sternlicht MD, Freiman CJ, Huaman JA, Barsky SH. The 25. Gusterson BA, Warburton MJ, Mitchell D, Kraft N, Hancock WW.
primary tumor is the primary source of metastasis in a human melanoma/ Invading squamous cell carcinoma can retain a basal lamina. Lab Invest
Scid model. Implications for the direct autocrine and paracrine epigenetic 1984; 51: 82–87.
regulation of the metastatic process. Int J Cancer 1996; 66: 151–158. 26. Sirtori C, Talamazzi F. Il carcinoma intraduttale della mammella non e’mai
2. Stephenson RA, Dinney CPN, Gohji K, Ordonez NG, Killion IJ, Fidler IJ. un carcinoma in situ. Tumori 1967; 53: 641–644.
Metastatic model for human prostate cancer using orthotropic implantation 27. Van de Vijver M, Peterse J, Mooi W, et al. Neu-protein overexpression in
in nude mice. J Natl Cancer Inst 1992; 84: 951–956. breast cancer. N Engl J Med 1988; 319: 1239–1245.
3. Weiss L. Metastatic inefficiency. Adv Cancer Res 1990; 54: 159–211. 28. Meyer JS. Cell kinetics of histologic variants of in situ breast carcinoma.
4. Holland R, Peterse JL, Millis RR, et al. Ductal carcinoma in situ: a proposal Breast Cancer Res Treat 1986; 7: 171–180.
for a new classification. Semin Diagn Pathol 1994; 11: 167–180. 29. Leygue E, Snell L, Hiller T, et al. Differential expression of psoriasin
5. Silverstein MJ, Poller DN, Waisman JR, et al. Prognostic classification of messenger RNA between in situ and invasive human breast carcinoma.
breast ductal carcinoma-in-situ. Lancet 1995; 345: 1154–1157. Cancer Res 1996; 56: 4606–4609.
6. Fu YS, Cheng L, Huang I. DNA ploidy analysis of cervical condyloma and 30. Sharkey FE, Greiner AS. Morphologic identity of primary tumor and
intraepithelial neoplasia in specimens obtained by punch biopsy. Anal Quant axillary metastases in breast carcinoma. Arch Pathol Lab Med 1985; 109:
Cytol Histol 1989; 11: 187–195. 256–259.
? 1997 John Wiley & Sons, Ltd. , . 183: 188–194 (1997)