Breast Cancer Heterogeneity:
`
`Evaluation of Clonality in Primary and Metastatic Lesions
`
`W. FRASER SYMMANS. MB. CHB. JIGUAN LIU, MD,
`
`DANIEL M. KNOWLES, MD, GIORGIO INGHIRAMI. MD
`
`Breast cancers often contain different clones of tumor cells.
`
`Attention to the cellular properties of breast cancer metastases may
`identify characteristics in primary tumors that are associated with
`metastasis. Such chmcteristies could include DNA content, cell pro—
`liferation, abnormal oncogene expression, or relative cell population
`(clonal dominance}. We examined DNA ploidy (image analysis}, pro-
`liferation index (proliferating cell nuclear antigen-l inununostain-
`ing), and expression of HmZ/m oncoprotein in 17 invasive breast
`cancer samples (316 primary tumor samples) and 82 corresponding
`regional metastases. In all samples the primary tumor was multiclonal
`(usually biclonal} by DNA ploidy analysis. In approximately 90% of
`metastatic DNA clones (30 of 34) the corresponding clone was identi-
`fied in a primary tumor sample representing 25% or more of the
`tumor cell population (significant clone}. A majority DNA clone
`(350% oftumor cell population) existed in 60% (21 of 36) of primary
`nunor samples and in 70% (60 of 82) of metastases (30% diploid v
`70% nondiploid in both groups). In approximately 50% of metastases
`(3'? of 82) an unexpected majority clone was identified (not a majority
`in any primary tumor sample} and the ratio of diploid to nondiploid
`clones also was 30% to 70%. However, in 80% of majority metastatic
`clones (46 of 60} that clone was a significant primary tumor clone.
`
`Considerable attention has been paid to the pre-
`dictive value of DNA ploidy, proliferation index, and
`Her—Z/ngu oncoprotein expression in breast cancer. It
`is believed that most breast carcinomas have a mono-
`
`clonal ploidy profile {single majority DNA peak), of
`which 25% to 35% are diploid and 65% are nondiploid
`(including ancuploid, hypodiploid, and tetraploid)."2
`A recent review of the flow cytometry literature reports
`the incidence of multiploidy in single tumor samples
`as 10% (range, 4% to 19%).
`In our laboratory we find
`that 8% of tumors are multiclonal for DNA index by
`image analysis of single tumor samples. However, there
`is heterogeneity of the DNA index in approximately
`25% to 40% of breast carcinomas when multiple re—
`gions of the primary tumor are examined by flow cytom-
`etry.% Higher rates of multiploidy (50% to 70%} have
`been reported when four or more samples are exam-
`ined.“‘7
`Animal studies of melanoma and breast carcinoma
`
`suggest that the relative proportions of difl'crcnt clones
`may relate to the metastatic process.”[5 It has been pos-
`tulated that the dynamic equilibrium of different clones
`
`
`From the Department of Pathology, College of Physicians and
`Surgeons of Columbia University, New York, NY. Accepted for publi-
`cation June 20, 1994.
`Presented in part at the $15!; annual meeting of the International
`Academy of Pathology, United States—Canadian Division, March
`1992.
`Address correspondence and reprint requests to Giorgio Inghir—
`ami, MD, Department of Pathology, New York University Medical
`Center, 560 First. Ave, New York, NY 10015.
`Copyright © 1995 by WB. Saunders Company
`0M6-8177/95/2GU2-Ufll255flO/0
`
`Proliferation index was quite variable in primary tumor samples and
`in corresponding metastases. Overexpmsion of Her-B/m oncopro-
`tein in the primary tumor of seven of 10 patients also was identified
`in all corresponding metastases in five of seven patients and in some
`metastases in two of seven patients. The metastases in three Her—Z/
`neat-negative patients were all negative. We conclude that (1) DNA
`clones are stable after metastasis, (2) clonal majorities in metastases
`reflect clones identified in primary tumors, (3) different metastatic
`clones from an individual tumor can establish clonal majorities, (4}
`neither diploid nor aneuploid cells have a metastatic advantage in
`breast cancer, (5) proliferation indices are heterogeneous, and (ti)
`overexpression of Hertz/net: is usually consistent between primary
`tumors and corresponding metastases. HUM PATHOL 26:210-
`216. Copyright © 1995 by W.B. Saunders Company
`Key words: breast, cancer, DNA, oncogene, ploidy, heterogeneity,
`proliferation, proliferating cell nuclear antigen, Her-Hm, Gerbil-2,
`metastatic, multiclonal, clones.
`Abbmn'atiom: PCNA—l, proliferating cell nuclear antigen—l; ABC,
`avidin biotin complex; DAB, diaminobenzidine hydrochloride; Ah,
`antibody; RT, room temperature; APAAP, alkaline phosphatase anti-
`alkaline phosphatase.
`
`within a tumor is eventually overcome by a biologically
`dominant clone with enhanced metastatic potentiallys
`Thus, a study of DNA ploidy in the metastases of
`multiclonal tumors provides an interesting model in
`which two significant and presumably metastatically
`competent clones compete for biological advantage.
`Quantitation of proliferating cell nuclear antigen-
`] {PCNA—l) Staining by image analysis correlates well
`with other techniques for assessing proliferative activ—
`ity.'6"8 Proliferating cell nuclear antigen—l is identified
`in nuclei during late 01, S, and G2 phases of the cell
`(311216.ng Furthermore, PCNA—l may be reliably used in
`paraffin—embedded tissue sections.'6"7'2"‘2' We chose to
`examine the variability in proliferation index within
`primary breast cancers and metastases. We also used
`the proliferation in these samples to confirm that non-
`diploid populations were separate clones rather than
`proliferating diploid cells.
`Overcxpression of Her-2/neu oncoprotein may con—
`fer a biological advantage on tumor cells. This onco-
`protein is a transmembrane receptor that has similar
`and perhaps tandem functions to epidermal growth fac-
`tor receptor.” Thus, overexpression of this protein may
`be associated with growth advantage and metastatic ad-
`vantage in tumor cells.22‘23 In studies with long clinical
`follow-up, overexpression of Her—2mm may be associ-
`ated with reduced overall and disease-free survival.22’24’25
`
`The prognostic significance may be greater in node-
`positive patien 6.25 However, the literature does not uni—
`formly support a prognostic role for Her—2/neu overex-
`prcssion in breast calico-1225‘27
`We evaluated clonal heterogeneity relating to tu—
`mor progression by using image analysis in 17 patients
`
`2'"
`
`Genentech 2107
`
`Hospira v. Genentech
`lPR2017-OO737
`
`Genentech 2107
`Hospira v. Genentech
`IPR2017-00737
`
`

`

`BREAST CANCER HETEROGENEITY (Symmons e1 0|)
`
`with breast carcinoma and 82 corresponding metasta-
`ses. Heterogeneity was frequent. in primary and meta—
`static tumor samples but consisted of relatively stable
`subcloncs. There was no metastatic advantage based on
`DNA content. Her—Z/mm gene expression was consis—
`tently maintained as a marker of clonality.
`
`MATERIALS AND METHODS
`
`Patients and Pathological Specimens
`
`We identified 17 patients with invasive breast carcinoma
`that was mttlticlonal by quantitative DNA ploidy analysis. A
`total of 36 samples of primary tumors was studied. All patients
`had metastatic tumor and a total of 82 regional metastases
`was available for study. Serial sections were obtained from
`paraffinembedded tissue that had been fixed in 10% form-
`lin or Bouin‘s solution.
`
`Feulgen Staining
`
`Deparaffinized 611m thick tissue sections (xylene fol-
`lowed by decreasing alcohol solutions) were rinsed in distilled
`water for 5 minutes and then stained using the Feulgen
`method (Quantitative DNA Staining Kit, Cell Analysis Systems
`[CA5] , lnc, Elmhurst, IL)?” Briefly, the sections were placed
`in 5 N HCL for 60 minutes to hydrolyze DNA. They were
`then stained with Feulgen stain for 1 hour and rinsed in three
`changes of “rinse solution" (Quantitative DNA Staining Kit)
`for l, 5, and 10 minutes, respectively, and Iinally in running
`distilled water for 5 minutes. After dehydration in 70% acid
`alcohol for 5 minutes and further dehydration in 100% etha—
`nol for 3 minutes, the slides were placed in xylene and then
`coverslipped.
`
`Immunohistoehemicol Methods
`
`Paraffin sections were stained with anti—PCNA—l anti—
`
`body using an avidin biotin cOmplex (ABC) method with di—
`aminobenzidine hydrochloride (DAB) chromagen. Briefly,
`sections were blacked for endogenous peroxidase activity
`(0.45% hydrogen peroxide in methanol for 15 minutes],
`coated with 10% horse serum (10 minutes), incubated with
`primary antibody [PCNA—l, P010 clone [Dako, (Earpinteria,
`CA], mouse monoclonal antibody [Ab] at 1:200 concentra-
`tion overnight at 4°C), secondary Ah (30 minutes at room
`temperature [RT]), ABC complex (30 minutes at RT), and
`then developed with freshly prepared DAB chromagen (3 to
`7 minutes at RT). Sections were washed in phosphate buffered
`saline between each step. The nuclei were counterstained
`with ethyl green in sodium acetate buffer. The sections were
`then dehydrated, cleared, and mounted.
`Her-Z/mrr oncoprotein (c—erbBr?) was detected by immu—
`nostaining with an alkaline phosphatase antialkaline phospha—
`tase (APAAP) method with red chromagen (CA5) after Feul—
`gen staining as described above. Normal goat serum
`“blocking reagent." (GAS) was applied for 20 minutes at RT,
`followed by primary Ab (Oncogene Science [Uniondale, NY],
`mouse monoclonal Ab, 1:200 concentration) overnight at
`4°C. Primary Ab was developed by an APAAP method as de—
`scribed prewiously.23""9
`
`puter was performed by measuring at least 20 control cells
`(rat hepatocytes) with a known nuclear mass (7.18 picograrns}
`from which the mode of the summed optical density per cell
`was obtained. The computer calculated a K value from this
`data by the linear equation: nuclear mass = K(EOD). This K
`value then was used by the system to compute the nuclear
`mass for each test cell and to report the modal value of the
`primary peak in picograms. The DNA index was obtained by
`dividing the modal value of the test cell by the DNA mass
`obtained from normal (diploid) adipocytes or histiocytes or
`the near-diploid lymphocytes collected in a dillerent channel.
`The proliferation index was determined by quantitation
`ofnuclear PCNA-l iminunostaining using image analysis. Soft-
`ware supplied with the CAS—200 was used to quan titate nuclear
`staining by determining the percentage of the optical density
`represented by positive tumor cell nuclear staining within
`the areas defined as nuclei. Thresholds were set for each
`
`microscopic field (400x) both for nuclear image (ethyl green
`stain) and for positive PCNA—l immunostaining. Window con—
`trols allowed tin—screen gating of the tumor areas within each
`microscopic field. The percentage of nuclear area staining
`positively was measured in 10 diITerent, unselected micro
`scopic fields and the mean was calculated.
`Her—Z/nm oncoprotein (c—afiB2] staining was quantitated
`using software supplied with the GAS—200. Briefly, two image
`sensing channels were used, one for the blue Feulgen stain
`(nuclei) and die odier for the red APAAP ilnmunostained
`Her—Ween oncoprotein (membrane and cytoplasm), Each cell
`measured had both parameters quantitated. A tumor cell line
`that overexpresses a constant amount of Her-2/neu oncopro—
`tein was used to calibrate the GAS-200 to 100% protein con—
`tent (1 picogram of [Her—2/nm]/picogram of DNA). By ini—
`tially quantitating the tumor cell DNA content in picograms
`per cell (see DNA quantitation above), the density of mem-
`brane and cytoplasmic Her—Z/nm oncoprotein staining (pica-
`gram of Ing—2/neuJ/picogram of DNA) could then be ex-
`pressed as an average value of Her-2km oncoprotein staining
`per cell. Window controls allowed on-screen gating of the
`tumor areas within each micrOscopic field. At least ll} micro-
`scopic fields (400X) were measured and the mean value was
`expressed as picograms of (Her—Z/rwu) per cell.
`
`Interpretation
`
`Each DNA histogram was independently analysed ac—
`cording to modal DNA indices. The proportion of cells in
`each DNA clonal population was calculated by assessing the
`contour and range 01' DNA indices for each peak in the histo-
`gram and determining the percentage of tumor cells con—
`
`A
`
`1 .7
`
`1 . 0
`
`B
`
`2 1
`
`_l
`a
`LU
`
`0.
`
`I .7
`
`Image Analysis
`
`DNA INDEX
`
`DNA INDEX
`
`All sections were studied microscopically and analysis was
`confined to areas of infiltrating, viable appearing carcinoma.
`Computerized image analysis (GAS-200, (3A5, Elmhurst, IL)
`was used to evaluate DNA content.”'29 Calibration of the com-
`
`FIGURE 1. Representative examples of DNA histograms of (A)
`o diploidz’nondiploid tumor and (B) a nondiploidlnondiploid
`tumor. Coefficients of variance tor DNA populations ranged
`from 10% to 24%.
`
`211
`
`

`

`HUMAN PATHOLOGY
`
`Volume 26. No. 2 (February l995)
`
`TABLE '1.
`
`Clonol Distribution in Primary and Metastatic Tumors (DNA Content)
`
`Patient
`No.
`
`Primary Tumor
`
`Metastases
`
`l
`2
`3
`4
`5
`6
`7
`3
`9
`10
`11
`12
`13
`
`14
`15
`16
`17
`
`[1.5, 1.0]; 1.4; 3): 1.0
`[1.11). 1.4, 1.6]; [1.3, 1.0]; 1.0
`2.1; 2x 1.5
`[1.5, 2.0:
`[1.8, 2.8]
`2.0; 3:; 1.0; 0.7; 5:: 0.8
`W; 2.1; 2.4; 2x 2.2
`[1.2, LID]; 2x 0.7
`1.0 _._.._
`[1.1NT}, 1.4]; [1.1D, 1.2, 1.4}; [1.113, 1.4}; [1.3, 1.11)]; 1.3; 3): 1.11)
`1.4
`27:13
`
`[1.0, 1.3]; [0.91), 1.4-. 1.8]
`[1.0, 1.5]; 0.7; 0.8
`[1.4, 2.0]; [2.0, 1.4]
`[1.6, 2.0}
`[1.0, 2.0]
`1.0; 0.8; 0.9N1)
`[1.5, 2.0, 2.3]; 1.0
`[0.8, 1.8]; 0.9ND
`[1.0, 1.5}; 1.3; 0.9D
`[1.0, 1.3, 1.6]
`[0.9D, 1.5]; [1.0, 1.4]
`{2.0,1.3};l1.0,1.3}
`{0.90, 1.4}; [1.0, 1.5, 1.9};
`[17,111]
`1.3; 0.90
`[1.6, 0.91)]
`2x 1.0; 3x 1.7; [1.7, 2.4]; [1.7, 2.9]; [1.3, 2.6]; [2.0, 1.6]: 1.4; 3x 1.6
`{1.1, 2.1}; {2.1, 1.7}; [2.0, r.o|;1.3
`0.7
`[1.7, 2.1}; 11.6, 0.3}
`
`{2.0, 2.6] 2.9
`
`2): [0.7, 1.0]; [0.8, 1.5]; 21: [1.0, 1.5]; [1.0, 2.0}; [1.5, 2.0. 3.0]; [1.5, 1.0}; 1.5; 1.6; Ill—.0; 3x E; 8:: 0::
`
`
`
`NOTE. DNA indices of significant (;2B%) populations are shown for all samples. DNA indices of majority (250%) populations are in
`bold face. Multiclonal samples are shown in parentheses. Underlined DNA indices are majority metastatic clones not identified as a significant
`clone in the primary tumor. D or NU indicates whether a near—diploid DNA index was interpreted as diploid or nondiploid before rounding
`the DNA index from two to one decimal places.
`Abbreviation: nx = n samples.
`
`tained within each distinct peak. Ploidy status was assigned to
`each clone as follows: diploid (DNA index 1.0 i 0.1) and
`nondiploid (DNA index < 0.90 or >110; Fig l). A significant
`population was defined as a population comprised of at least
`25% of die tumor cells. A nondiploid clone was defined as a
`significant nondiploid population it the percentage of tumor
`cells in that clone was greater than double the proliferation
`index measured by PCNA-l
`immunostaining and was com-
`piised of at least 25% of the tumor cell population. A majority
`population was defined as a population comprised of at least
`50% of the tumor cells in that sample. An unexpected major-
`ity population was defined as a majority DNA clone in a metas-
`tasis that was not present as a majority population in any of
`the primary tumor samples from that patient. if two DNA
`peaks were within a range of 10% (0.1 for hypodiploid peaks)
`of each other, they were considered to represent the same
`clone. Direct comparison Of all the DNA profiles from each
`patient‘s samples was needed to reliably distinguish the dill'er—
`ent clones within these histograms.
`Her—2/m oncoprotein staining greater than 0.1 pg/ cell
`was considered to represent overexpression of the protein
`and was termed positive. The upper limit of normal Her—2/
`mm oncoprotein expression has been described as 10% of the
`positive control cells, ie, 0.1 pig/cell?”
`
`RESULTS
`
`The indices of significant and majority DNA
`clones, proliferation indices, and expression of Her-2/
`HEB. oncoprotein for all samples are presented in Tables
`1 through 4. At least one tumor sample had two clones
`represented as significant. tumor cell populations within
`a single sample in 16 of 17 patients. In the remaining
`case (patient no. 6) the two clones were each a majority
`population in different samples from the primary tu—
`mor (Table 1).
`There were 48 significant DNA clones in 36 pri—
`212
`
`mary tumors. We found 60% of these clones (29 of 48)
`to be metastatic as a significant population. Conversely,
`in 82 metastases there were 34 significant clones. Ap-
`proximately 90% of the metastatic clones (30 of 34)
`were identified as a significant population in the pri-
`mary tumor. A majority DNA clone was identified in
`approximately 50% of primary tumor samples (21 of
`36), of which 29% were diploid and 71% were nondip—
`loid (Table 2). Similar results were obtained for the
`metastases (Table 3). Considering the 60 metastases
`with a majority DNA clone, the metastatic cleric was
`unexpected in 62% (37 of 60), but in approximately
`80% (46 of 60} it was present as a significant clone in
`the primary tumor (Table 1).
`A combination of diploid and nondiploid DNA
`clones was identified in the primary tumor of 12 of
`17 patients. A majority DNA clone was identified in
`approximately 60% of primary tumor samples (17 of
`28), of which 35% were diploid and 65% were nondip-
`loid (Table 2). Similar results were obtained for the
`metastases as for the primary samples and for the metas-
`tases from the en tire study p0pulation (Table 3). In the
`52 metastases with a clonal majority the metastatic clone
`was unexpected in 32 (62%), but in 75% of these (39
`of 52} it was present as a significant population in at
`least one sample of the primary tumor. One case (pa-
`tient no. 13) accounted for 23 of 73 metastases (32%)
`in the diploid/nondiploid group and contained 13 un—
`expected metastatic majority clones, of which 12 repre—
`sented a hypodiploid clone not identified in the pri-
`mary tumor (Table 2). This patient may have created
`a bias in these results. We then excluded patient 13 and
`found that in 25 primary tumor samples there were 16
`majority clones (64%): five (31%) were diploid and 12
`(69%) were nondiploid. In 50 metastases there were 38
`(76%) majority clones (39% diploid and 61% nondip—
`
`

`

`BREAST CANCER HETEROGENEITY (Symmons et ol)
`
`patients had overexpression of Her—2/neu in the primary
`tumor {one to four samples each). Overexpression of
`Her—2/neu was present in all metastases studied (five of
`seven patients) or in a proportion of them (patients
`no. 1 and 13; Table 4). Her—Z/nett was not. overexpressed
`in all metastases from those primary tumors that were
`negative for antigen (three patients).
`
`DISCUSSION
`
`Our findings indicate that the DNA clones repre—
`sented in the metastases of multiclonal breast carcino-
`
`mas reflect the proportion of clones in the primary
`tumors. We found that 90% of significant. metastatic
`clones were present. as a significant. clone in a primary
`tumor sample. The DNA indices of significant DNA
`clones in primary tumors and their metastases were sim—
`ilar (Table 1). Thus, the same DNA clones in the pri—
`mary tumor are those that metastasize.
`The ratio of diploid to nondiploid majority clones
`was approximately 30% to 70% in primary tumor sam—
`ples and in their metastases. This ratio was the same
`irrespective of whether the metastatic clone was a major-
`ity population in the primary tumor samples {Table
`3). Thus, both ploidy types have equivalent metastatic
`ability. This is in contrast to the results of Alam et alf”
`who compared DNA indices of single tumor samples
`with single rrletastases in 12 patients. They reported
`four multiploid tumors in which all the corresponding
`metastases were composed of the same nondiploid
`clone as in the primary sample? Their conclusion that.
`nondiploid DNA subpopulations may have a metastatic
`advantage is not confirmed by our data. Auer et 3132
`determined that ploidy was similar in primary tumors
`and their metastases. We confirm that the DNA indices
`
`are similar; however, we differ with respect to the pro-
`portions of DNA populations and the proliferation indi—
`ces, which our reSults show to be more variable.32 Other
`investigators have shown variation of ploidy status be-
`tWeen primary tumors and metastases inaapproximately
`10% to 25% of cases.33"“ Bonsing et a1
`studied all
`
`
`TABLE 3. Summary of Metastatic DNA Clones
`
` %
`
`Frequency
`
`9i;
`
`%
`Diploid Nondiploid
`
`All patients
`Majority clone
`Unexpected majority
`Majority, no 1° clone
`D/ ND patients
`31
`71
`52/73
`Majority clone
`31
`44
`32/73
`Unexpected majority
`
`Majority, no 1° clone
`13/73
`18
`0
`
`60/82
`37/82
`14/82
`
`73
`45
`17
`
`28
`30
`7
`
`72
`70
`93
`
`69
`69
`100
`
`NOTE. Majority clones comprise 50% or more of a. tumor cell
`population. Unexpected majority clones were not found as a majority
`clone in any corresponding primary tumor samples. Majmity, no 1"
`clone indicates majority metastatic clones for which no corresponding
`significant clone [225% of tumor cell population) was found in the
`primary tumor. D/ND patients are the IQ patients with diploid and
`nondipioid DNA clones in the primary tumor.
`Abbreviations: 1'), diploid; ND, nondiploid.
`
`TABLE 2. Majority DNA Clones
`Metastases
`
`Primary Tumor
`
`Unexpected
`
`No.
`No.
`No.
`No.
`No.
`No.
`No. of
`Patient
`
`No.
`Samples
`D
`ND
`D
`ND
`D
`ND
`No.
`
`l
`‘2
`5
`6
`7
`9
`10
`l l
`12
`l 3
`14
`15
`
`Diploid/Nondiploid Tumors
`0
`t}
`5
`4
`t]
`3
`3
`2
`(l
`U
`1
`t}
`l
`2
`l [1
`3
`l
`0
`5
`0
`l
`1
`1
`1
`0
`0
`7
`4
`0
`l
`1
`l}
`0
`2
`2
`[1
`1
`U
`23
`1
`1
`0
`2
`1
`1
`2
`13
`0
`
`l
`0
`{J
`7
`5
`0
`2
`l
`2
`l 3
`0
`5
`
`2
`3
`l
`3
`2
`3
`l
`3
`2
`3
`1
`4
`
`4
`2
`U
`[l
`0
`0
`4
`l]
`[1
`U
`U
`0
`
`l
`0
`U
`l
`5
`{J
`2
`{l
`t]
`1 3
`0
`0
`
`Nondiploid/ Nondiploid Tumors
`l
`0
`3
`0
`l
`3
`0
`2
`3
`0
`0
`0
`0
`0
`l
`0
`l
`4
`2
`l
`2
`0
`i
`3
`1
`2
`8
`0
`0
`l.
`0
`2
`l
`0
`2
`16
`
`
`l17 l 0 0 l 0 l 0
`
`
`
`
`
`
`
`NOTE. The number of diploid or nondiploid majority tumor
`cell populations in the primary tumor samples and metastases for
`each patient were compared with the total number of samples and
`metastases for each patient. The unexpected column indicates the
`number of diploid and nondipioid majority tumor cell populations
`in the metastases of which no majority of that DNA clone was identi—
`fied in any sample from Lite primary tumor. The patients were divided
`into two groups depending on whether the different DNA cl0nes in
`the primaiy tumor were diploid /nondiploid or all nondiploid.
`Abbreviations: D, diploid; ND, nondiploid.
`
`loid). [n the 38 metastases with a clonal majority the
`metastatic clone was unexpected in .19 (50%), but in
`97% of these (37 of 38) it was present as a significant
`population in at least one sample of the primary tumor.
`Therefore, excluding patient 13 did not. markedly
`change our results. Approximately 30% of majority
`DNA clones in both primary and metastatic tumor sam—
`ples (including unexpected majority clones) are diploid
`and approximately 70% are nondiploid.
`A combination ofdifferent nondiploid DNA clones
`was identified in the primary tumor of five of 17 pa—
`tients. In the eight primary tumor samples there were
`four majority clones (50%). In nine metastases there
`were eight majority clones (90%; all nondiploid) and
`one unexpected majority clone. However,
`this meta-
`static clone was present as a significant pepulation in
`the primary tumor.
`There was some variability in proliferation indices
`within the samples of each primary tumor. However,
`no observable difference between the proliferation1n—
`dices of the primary tumor samples and their corre—
`sponding metastases was found (Table 4). This results
`partly from the wide range of proliferation indices be—
`tween samples and among metastases of the same tu—
`IIlOr.
`
`Her-2/neu oncoprotein was studied in both primary
`and metastatic tumors in 10 patients (Table 4). Seven
`
`213
`
`

`

`HUMAN PATHOLOGY
`
`Volume 26. No. 2 (February 1995)
`
`
`TABLE 4. Clonal Distribution in Primary and Metastatic Tumors (Proliferotlng Cell Nuclear Antigen and Her—2/neu)
`Metastases
` Primary Tumor
`% PCNA
`
`% PCNA
`
`Patient
`
`
`Range
`Her2/netr
`
`(3):) +; (2x) —
`(Bx) +
`
`8—11
`9—20
`l—5
`
`0-7
`
`4—6
`
`14-10
`
`(3x) +
`
`5—51
`
`11—15
`3—9
`
`Range
`
`18—19
`s] 5
`4—] 7
`
`3-28
`
`4—7
`13-14
`
`2&4?
`8—20
`14—44
`
`11
`
`‘2
`s
`‘2
`l
`
`3
`1
`2
`3
`1
`2
`2
`3
`
`1
`4
`2
`1
`
`No.
`
`Mean
`
`18.5
`7.7
`10.5
`7
`
`12.3
`l l
`5.5
`I 3.7
`19
`35
`14
`27.3
`
`17
`9.25
`5
`19
`
`1
`2
`3
`4
`
`55
`
`7
`3
`9
`10
`11
`12
`13
`
`14
`15
`16
`17
`
`HerZ/mu
`
`Mean
`
`+
`(2x) +
`
`+
`
`9.8
`14.5
`3
`lfi
`1i}
`3
`
`It
`
`4
`2
`2
`1
`1
`9
`
`+
`{2x} —
`—
`(Ex) +
`
`5
`4
`22.7
`24
`2i
`21.6
`
`2
`1
`7
`1
`1
`22
`
`(llx) +;
`(12x) —
`—
`(2x) —
`2
`is
`(13):) +
`13
`4.8
`(43:) +
`315
`+
`1
`7
`(2x) +
`3-7
`
`
`
`2 l+ +
`
`NOTE. Summary of proliferation index (% PCNA immunostaining by nuclear area) and Her—2/neu expression (920.1 pg/ccll is considered
`positive). Proliferating cell nuclear antigen is expressed as the mean. range, and number (n) of primary tumor samples or metastases studied.
`Abbreviations: PCNA, proliferating cell nuclear antigen; rtx = r1 samples; +, positive; —, negative.
`
`available metastases from a similar number of patients.
`They found that 80% of DNA stemlines (clones) from
`primary tumors were identified in metastases (we found
`60%) and 64% of DNA stcmlines in metastases were
`identified in the primary tumors {we found 90%).35
`Our more frequent finding of metastatic clones in the
`primary tumors may reflect our sample selection of
`multiclonal tumors as well as a lower threshold for iden-
`
`tification of a significant clone (325%). This lower
`threshold can be afforded by the specificity with which
`image analysis can measure only tumor cells.
`The majority of published studies show that DNA
`ploidy does not predict nodal status in breast cancer.“638
`Because many breast cancers contain multiple DNA
`clones, and we describe equivalent metastatic propor-
`tions of diploid and nondiploid clones in multiclonal
`cancers, we would not expect single measurements of
`DNA ploidy in breast tumors to accurately predict nodal
`status.” Furthermore, DNA ploidy probably is not a
`useful single prognostic indicator in node-positive
`breast cancer? 9'40
`DNA index is a crude measure of true DNA con-
`
`tent. Our range of diploid values (1.0 i 0.1) would
`include cells that had gained or lost up to 10% of their
`DNA. Thus, it is possible that diploid clones harbor
`cytogenetic abnormalities not detectable by ploidy anal—
`ysis. Similarly, aneuploid clones with the same DNA
`index could contain very different karyotypic abnormal-
`ities. Such cytogenetic changes could result in the emer—
`gence of new subcloncs that are beyond the resolution
`of ploidy analysis. Thus, cytogenetic differences could
`exist between primary and metastatic tumor cell popula—
`tions with the same DNA index. Cytogcnctic analysis of
`these patients’ tumors was not performed. Also, some
`of the diploid clones may actually be near diploid. How‘
`
`ever, this criticism ignores the normal variance of this
`technique in paraffin—embedded tissues in which nor-
`mal stromal cells with known diploid status have a DNA
`index of 1.0 i 0.1. We used broad criteria to study
`the relative proportions of different clones (225% and
`250%). Although sensitivity to detect small differences
`in DNA clonal populations was lost, there was less possi—
`bility of misinterpreting small variations in the percent-
`age of DNA clones. It might be argued that the biphasic
`DNA profiles in these tumors do not represent two
`clones of tumor cells but rather are different cell cycle
`phases of a single clone. However, the consistent prescr-
`vation of DNA indices in primary tumor samples and
`metastases as well as the absence of higher proliferation
`in the nondiploid samples is strong evidence for the
`presence of stable DNA clones in these tumors.
`The metastases from each single tumor are hetero-
`geneous by DNA ploidy (Tables 1 and 2), suggesting
`that. a single DNA clone does not have overall metastatic
`dominance in the metastases. The variability in DNA
`profiles of primary tumor samples may contradict an
`argument for clonal dominance in the primary tu-
`mors.”'15 However, the same clones from the primary
`tumor usually are found in corresponding metastases,
`there are similar rates of majority DNA clones in pri—
`mary (60%) and metastatic (70%) samples, and 80% of
`majority metastatic clones are identified as a significant
`population in at least one primary tumor sample. Thus,
`stable DNA clones exist as significant populations in
`the primary and metastatic tumors and new subclones
`infrequently emerge in the metastases. Although this is
`not strictly consistent. with Kerbel’s” theory of clonal
`dominance, these findings do support some form of
`equilibrium or codominance of DNA clones in breast
`C31]CCI’.H
`
`214
`
`

`

`BREAST CANCER HETEROGENEiTY (Symmons at of)
`
`The proliferative activity data show variation of pro—
`liferation index among samples of the same tumor,
`among metastases, and within both diploid and nondip-
`loid samples. Significant correlations between S—phase
`fraction in single tumor samples and single metastases
`have been observed, although there was wide variance
`in these seatterplots.33'41 We predict that analysis of mul—
`tiple samples from each grOup would further increase
`this variance. lmmunohistochemical staining with the
`PC10 clone of PCNA is not influenced by delayed fixa—
`tion”; however, it is diminished after
`rolongcd fixa—
`tion in formalin for more than 3 hours‘1 and more than
`24 hours in other studiesmg' The majority of samples
`for each patient were obtained from resection speci—
`mens and fixed for the same length of time. Thus, the
`variance in proliferative activity between samples should
`not be attributed to different fixation times.
`
`Her-B/neu overexpression was remarkably consis—
`tent between primary tumors and metastases despite
`considerable variation in DNA ploidy, and is most likely
`independent of ploidy. Although one study suggested
`that
`tetraploid tumors overexpress Her-Z/neu more
`frequently than other ploid
`roups,29 subsequent stud-
`ies were not confirmatory?“ ‘30 Data from frozen and
`paraffinembedded tissues are probably similar when
`Harri/mu immunostaining is quantitated by image anal—
`ysis.” Other investigators have described high concor—
`dance between Her-Z/neu oncoprotein overexpression
`{and epidermal growth factor receptor expression) in
`breast carcinomas and in their metastases
`7; however,
`not all metastases from Her—2/neu~positive tumors are
`positive and some metastases from Her—Z/nsu—negative
`tumors are positive.“ We found that most metastases
`were Her—2/nert p0sitive if the primary tumor was posi—
`tive. Metastasis may be independent of Her—Z/neu ex-
`pression because the Her-2/nmmnegalivc primary tu-
`mors had Her—2/neu—negative metastases.
`We have not identified a “metastatic phenotype"
`with reSpect to DNA content, proliferation index, or
`Her-2/neu expression. A comprehensive attempt to char—
`acterize such a phenotype (if one exists} would need
`to consider the multifactorial nature of tumor cell biol—
`
`ogy as it relates to tumor growth, clonal interactions,
`vascularization,
`invasion, endothelial adhesion, meta—
`static colonization, and the interactions of tumor cells
`with host immune and stromal cells. Two recent studies
`
`described conservation of genetic characteristics be-
`tween primary and metastatic breast cancer cells, albeit
`in small patient numbers. Allelotype (described by DNA
`hybridizations using 31 different. probes) and loss of
`heterozygosity (DNA hybridizations with 12 different
`probes} showed the same genetic defects in samples of
`primary tumors and metastases in all patierltsfw'49 The
`stability of allelotype occurred despite heterogeneity of
`DNA index within the tumors.“ This may explain why
`we identified consistency of Her—Z/ne'u overexpression
`but not of DNA ploidy or proliferation index.
`We have shown that these tumors are composed of
`different, relatively stable DNA clones that can behave
`independently as metastatic populations. No metastatic
`advantage to diploid or nondiploid DNA clones was
`found.
`
`Acknnwiedgment. We thank A. Ward for excellent techni-
`cal assistance.
`
`REFERENCES
`
`l. Visschcr DW, Zariro Rj, Greenawald RA. et al: Prognostic sig-
`nificance of morphologic parameters and flow cytometric DNA analy—
`sis irr carcinoma of the breast. Pathol Annu 25:171-210, 1990
`2. Frierson HF: Ploidy analysis and S—phase fraction determina-
`tion by flow cytornetry of invasive adenocarcinomas of the breast. Am
`J Surg Pathol rszsssssr', 1991
`3. Kate T: Letter to the editor. Cylomctry 12:155, 1991
`4. Meyeer, Wittlifi'jL: Regional heterogeneity in breast carci-
`noma: Thymidine labeliing index, steroid hormone receptors, and
`DNA ploidy. Int] Cancer 47:21.3220, 1991
`5. I'litchcock A, Ellis 10, Robertson JFR, et al: An observation
`of DNA ploidy, histologir. grade and immunor‘eaclivity for tumors
`related antigens in primary and metastatic breast carcinoma] Pathol
`1592129134, 1989
`6. Bcerman II, Smit VTHBM, Kluin PM, et al: Flow cytometric
`analysis of DNA stcmline heterogeneity in primary and metastatic
`breast cancer. Cytomelry 12:147—154, 1991
`7. P'uhrjE, Frye A, Kattine AA, et al: Flow cytometric determina-
`tion of breast tumor heterogeneity. Cancer 67214014405, 1991
`8. Smroselsky A, Pafliak S, Fidler 1]: Changes in clonal composi—
`tion during in vivo growth of mixed subpopulations derived from the
`murine 11—1735 melan