Expression of topoisomerase IIalpha is associated with rapid cell proliferation, aneuploidy, and c-erbB2 overexpression in breast cancer.

Abstract

The role of molecular markers predicting the response to cytotoxic chemotherapy is not established. A potential predictive factor, topoisomerase IIalpha, is a target for certain cytotoxic drugs, and its concentration has been shown to correlate with chemosensitivity in vitro. We evaluated expression of topo IIalpha immunohistochemically in 230 breast cancer samples and studied its association with known clinicopathological factors and factors previously shown to predict response to cytotoxic drugs. Topo IIalpha protein expression was found in 0.6 to 39.4% (10.6 +/- 7.9%, mean +/- SD) of breast carcinoma cells, whereas expression was undetectable in nonmalignant breast epithelium. Topo IIalpha protein expression correlated well with semi-quantitative mRNA in situ hybridization (P = 0.007). A significant association was found between the proportion of topo-IIalpha-positive cells and low estrogen and progesterone receptor content (P<0.0001), high grade (P<0.0001), DNA aneuploidy (P=0.003), and c-erbB-2 oncoprotein overexpression (P<0.0001). Topo IIalpha expression was not associated with clinical variables, such as age of the patient, primary tumor size, or axillary nodal status. A highly significant linear correlation was found between topo IIalpha and tumor proliferation rate (S-phase fraction, r=0.46; P<0.0001). Because hormone receptors, grade, and ploidy are associated with tumor proliferation rate, topo IIalpha expression was adjusted for S-phase fraction to reveal the proliferation-independent clinopathological associations. According to the analysis of co-variance, only aneuploidy (P=0.0003) and c-erb-2 overexpression (P=0.01) were associated with proliferation-adjusted expression of topo IIalpha. In conclusion, the close association of Topo IIalpha with other potential predictive factors (tumor proliferation rate, c-erbB-2 oncoprotein) suggests that topo IIalpha, having a defined role as a target for cytotoxic drugs, may be a valuable predictor of response to chemotherapy.

Full text

Page 1

American Joiirnal ofPathology, Vol. 148, No. 6, June 1996
Copyright © AmericanSocietyjbrInvestigative Pathology
Expression of Topoisomerase Ila Is Associated
with Rapid Cell Proliferation, Aneuploidy, and
c-erbB2 Overexpression in Breast Cancer
Tero A. H. Jarvinen,* Juha Kononen,* Markku
Pelto-Huikko,t and Jorma Isola*
From the Laboratory ofCancer Genetics,* Tampere
University Hospital and Instituite ofMedical Technology,
and the Medical School,t University of Tampere,
Tampere, Finland
The role of molecular markers predicting the
response to cytotoxic chemotherapy is not estab-
lished. A potentialpredictivefactor, topoisomer-
ase Ha (topo Ha), is a targetfor certain cyto-
toxic drugs, and its concentration has been
shown to correlate with chemosensitivity
vitra We evaluated expression oftopoHa immu-
nohistochemically in 230 breast cancer samples
and studied its association with known clinico-
pathological factors and factors previously
shown to predict response to cytotoxic drugs.
Topo Haprotein expression wasfound in 0.6 to
39.4% (10.6 + 7.9%, mean ± SD) ofbreast carci-
noma ceUs, whereas expression was undetect-
able in nonmalignant breast epithelium. Topo IIa
protein expression correlated weUl with semi-
quantitative mRNA in situ hybridization (P =
0.007). A significant association was found be-
tween the proportion of topo-IIa-positive ceUs
and low estrogen andprogesterone receptor con-
tent (P < 0.0001), high grade (P < 0.0001), DNA
aneuploidy (P = 0.003), and c-erbB-2 oncopro-
tein overexpression (P < 0.0001). Topo Ha ex-
pression was not associated with clinical vari-
ables, such as age ofthepatient, primary tumor
size, oraxiUary nodal status. A highly significant
linear correlation was found between topo Ha
and tumorproliferation rate (S-phasefraction, r
=0.46; P < 0.0001). Because hormone receptors,
grade, and ploidy are associated with tumor
prolyferationrate, topo Ha expression was ad-
justedfor S-phasefraction to reveal theprolifer-
ation-independent
clinopathological
tions. According to the analysis of co-variance,
in
associa-
only aneuploidy (P = 0.0003) and c-erbB-2 over-
expression (P = 0.01) were associated with pro-
liferation-adjusted expression of topo Ha In
conclusion, the close association of Topo Ha
with other potential predictive factors (tumor
proliferation rate, c-erbB-2 oncoprotein) sug-
gests that topo Ha, having a defined role as a
target for cytotoxic drugs, may be a valuable
predictor of response to chemotherapy.
Pathol 1996, 148:2073-2082)
(AmJ
There are no commonly accepted markers that could
be used to predict response to cytotoxic chemother-
apy in breast cancer patients. Response to endocrine
therapy can be predicted by hormone receptor analy-
ses,1 but so far, prediction of response to cytotoxic
chemotherapy lacks reliable predictive markers. These
factors would be of utmost utility in the clinical man-
agement of breast cancer patients. Predictive factors
should not only predict response to chemotherapy in
general but should also be able to aid in the selection
between different forms of chemotherapy. With the in-
creased use of doxorubicin-containing regimens in-
stead of conventional cyclophosphamide-methotrex-
ane-fluorouracil,
theselection
regimens has become increasingly important. Topoi-
somerase (topo) Ila is one of the enzymes suggested
as a clinically relevant predictive factor.2
DNA topo 11 is a eukaryotic homodimeric enzyme
that exists in two isoforms in human cells; the 170-kd
topo Ila and the 180-kd topo Il3.3-8 These two iso-
forms of topo 11 share considerable homology but are
products of different genes located on chromo-
somes 17q and 3p, respectively. Topo Ila is a key
between
different
Supported by the Finnish Academy of Sciences, Finnish Cancer
Society, Sigrid Juselius Foundation, and Tampere Hospital Founda-
tion.
Accepted for publication January 31, 1996.
Address reprint requests to Dr. Jorma Isola, University of Tam-
pere, Institute of Medical Technology, Laboratory of Cancer Genet-
ics, PO Box 607, FIN-33101, Tampere, Finland.
2073

Page 2

2074
AJP June 1996, Vol. 148, No. 6
Jarvinen et al
enzyme in DNA metabolism, having a central key
role in DNA replication.3`8 Topo
generating and then resealing double-stranded DNA
breaks, which are necessary for segregation of chro-
mosomes at the end of mitosis.6'7 Although recent
reports indicate that topo lII, may have a physiolog-
ically relevant function as an integral component of
nuclear matrix and nucleoli9'10 and that it may play a
more significant role in mediating drug resistance
than has been previously appreciated,9-12 the func-
tion of topo 11p is still less clearly defined than that of
topo Ila.
Topo Ila is a target for a wide variety of structurally
diverse cytotoxic drugs, including some of the most
important anticancer drugs such as doxorubicin, m-
AMSA, mitoxantrone, and etoposide.'8 Drugs tar-
geted against topo 11 trap the enzyme in a so-called
cleavable complex, thereby inhibiting
function.4 This results in the accumulation of double-
stranded DNA breaks, which are lethal to the cell.4 It
has been shown that sensitivity to topo Il-inhibiting
drugs is correlated with expression to topo
vitro.13'14 This forms a hypothesis to test whether
assays of topo Ila could be used in a clinical setting
to predict a patient's response to chemotherapy.
Studies published so far suggest that a favorable
response to cytotoxic chemotherapy is associated
with a high tumor proliferation rate.15-18 A recent
study indicated that overexpression of c-erbB-2 on-
coprotein may be an even stronger predictor of re-
sponse to chemotherapy containing the topo 11 inhib-
itor doxorubicin.2 The biological background of this
finding is unknown, because the c-erbB-2 proto-on-
cogene encodes for a transmembrane growth factor
receptor, with no reported interaction with cytotoxic
drugs. Because topo Ila is located very close to the
c-erbB-2 oncogene at chromosome 17q12,
been speculated whether there
causing aberrant co-expression of these two pro-
teins.2,19-21 At the genomic DNA level, both amplifi-
cation19'20 and deletion21 of the topo Ia gene have
been described in association with c-erbB-2 amplifi-
cation. Correlation between topo Ila gene copy num-
ber aberrations and topo Ila protein expression are
not known.19-21
So far, only rather limited data are available on the
activity and expression of topo Ila in human neo-
plasms.22 26 Previous studies using antibodies to
topo 11 have confirmed that immunohistochemically
detectable levels of topo 11 are expressed in virtually
all breast cancer samples.25'26 In a pilot study of 63
invasive ductal breast cancers, Hellemans et a126
found a relationship between high expression of topo
Ila and high tumor grade, large tumor size, nodal
Ila acts first by
its proper
Ila in
it has
is a mechanism
status, and the distant metastases at the time of
diagnosis.
In this study we optimized an immunohistochem-
ical method for analysis of topo Ila expression from
frozen tissue sections and analyzed topo Ila expres-
sion in 230 primary invasive breast cancers. The
topo Ila expression was correlated with known clino-
copathological
indicators
among which were tumor proliferation rate and c-
erbB-2, the potential predictors of chemosensitivity.
of malignant
potential,
Materials and Methods
Patients and Tumors
We studied surgical biopsy specimens from a set of
230 randomly selected female breast cancer pa-
tients, whose tumor samples were sent for steroid
hormone receptor analysis to the Laboratory of Can-
cer Genetics, Tampere University Hospital. Entered
in this study were tumor samples from invasive, pri-
mary breast cancers. The median age of the patients
was 61 years (range, 28 to 89). Patients did not
receive any preoperative chemotherapy or endo-
crine therapy.
Tumor samples were snap-frozen in OCT tissue-
embedding medium (Tissue-Tek, Miles Laboratories,
Naperville, IL) within 15 minutes of removal. Cryostat
sections were cut and stained for rapid intra-opera-
tive histopathological diagnosis with routine meth-
ods. Extra sections (5 to 7 ,um) were cut and stored
air-tight at -700C until used in immunocytochemistry
and mRNA in situ hybridization. Histopathological
grading was performed according to the Bloom and
Richardson system using routine formalin-fixed, par-
affin-embedded sections, respectively.27
Immunohistochemistry of TopoI/a
Frozen sections were fixed with Zamboni's fluid for
15 minutes, washed with phosphate-buffered saline
(PBS), and incubated with 1.0% bovine serum albu-
min (BSA)/PBS for blocking nonspecific antibody
binding. A rabbit polyclonal antibody (TopoGEN, Co-
lumbus, OH) raised against a synthetic peptide of
the carboxyl-terminal region of human topo
incubated overnight (40C) at a dilution of 1:1000
(from manufacturer's suggested stock solution). A
standard avidin-biotin-peroxidase complex (ABC)
technique (Vectastain
Burlingame, CA) was used for visualization with dia-
minobenzidine as a chromogen. The diaminobenzi-
dine reaction was intensified with the methenamine
silver method as described elsewhere.28 Sections
11 was
Elite, Vector Laboratories,

Page 3

Topo Ila Expression in Breast Cancer
AJPJune 1996, Vol. 148, No. 6
2075
were then counterstained
mounted. Placenta was used as a positive control.
Conventional immunohistochemical controls (dilution
series of primary antibody and omission of primary or
secondary antibody or chromogenic substrate) indi-
cated that immunostaining was specific for topo Ila.
The percentage of positively stained carcinoma cells
(topo Ila score) was evaluated using a 20x objective
in a blind fashion, ie, unaware of estrogen receptor
(ER), progesterone receptor (PR), or c-erbB-2 con-
tent or other data. In the beginning, the whole histo-
logical section was examined under the light micro-
scope, and from morphologically well preserved
areas the highest topo
lected for analysis. From 400 to 1000 (average, 600)
carcinoma cells were counted to obtain the topo Ila
score. Correlation of topo Ila scores between two
observers was 0.960 (n = 10). Intra-observer repro-
ducibility of topo Ila scores was 0.867 (n = 15).
ERs and PRs were immunostained using Abbott's
ER-ICA and PR-ICA kits (Abbott Laboratories, Chi-
cago, IL). Overexpression of the c-erbB-2 protein
was detected immunohistochemically using a mono-
clonal antibody (NCL-BII, Novocastra Labs, New-
castle, UK).2021 The c-erbB-2, ER-ICA, and PR-ICA
antibodies were visualized as explained above for
topo Ila. Only intense membranous immunostaining
present in a majority of cells was taken to represent
overexpression of the c-erbB-2 protein, as
been shown to yield the best prognostic associa-
tions.29,30
with hematoxylin and
Ila staining region was se-
it has
Western Blot
Exponentially growing, BT 474 human breast cancer
cells were used for Western blots. Sodium dodecyl
sulfate polyacrylamide gel electrophoresis (SDS-
PAGE) was performed by the method of Laemmli31
with some modifications. Total cell lysates were pre-
pared by placing cultured cells directly into SDS
sample buffer (10% SDS, 5% 2-mercaptoethanol,
62.5 mmol/L Tris-HCI, pH 6.8), subjected to 8.5%
SDS-PAGE gels, and finally transferred to nitrocellu-
lose membranes. Immunostaining was carried out
using topo Ila as a primary antibody overnight at 40C
(concentration 1:10,000). To detect topo Ila, perox-
idase-conjugated swine anti-rabbit IgG antibody
(concentration 1:1000) (Dakopatts, Glostrup, Den-
mark) was incubated with blotting membranes for 60
minutes at room temperature. The peroxidase-cata-
lyzed reaction was visualized with enhanced chemi-
luminescence (Amersham Life Sciences, Arlington
Heights, IL). Molecular weight marker proteins con-
tained myosin (220 kd), phosphorylase b (97.4 kd),
and BSA (66 kd; Amersham Life Sciences, Arlington
Heights, IL).
Double Immunofluorescence Staining with
Topo /la and Ki-S1
Ki-Sl is an IgG mouse monoclonal antibody gener-
ated by immunizing BALB/c mice with crude nuclear
extracts from the human lymphoma cell line U937.32
As Ki-Sl has recently been shown to recognize hu-
man topo Ila,33 we double immunostained breast
cancer samples with Ki-Sl (generous gift from Prof.
H. Kreipe, University of Wurzburg, Wurzburg, Ger-
many) and anti-topo-lla to further demonstrate the
specificity
Briefly, frozen sections were fixed with Zamboni's
fluid for 15 minutes, washed with PBS, and incu-
bated with 1.0% BSA-PBS for blocking nonspecific
antibody binding. Topo
10,000) were incubated overnight in a cocktail (40C).
Topo Ila polyclonal antibody was detected by fluo-
rescein-isothiocyanate-conjugated goat anti-rabbit
IgG antibody (1:50; Jackson ImmunoResearch Lab-
oratories, West Grove, PA), whereas Ki-Sl was visu-
alized by biotinylated anti-mouse IgG antibody (1:
400; Vector Laboratories) followed by rhodamine-
avidin (1:200; Vector Laboratories). Sections were
then mounted in Vectashield A (Vector Laboratories)
and were analyzed with an Olympus B x 50 epiflu-
orescence microscope equipped with a 1Ox UP-
lanFI objective (Olympus Optical, Tokyo, Japan).
Two separate images (matching green fluorescein
isothiocyanate fluorescence and red rhodamine flu-
orescence) were captured with a scientific cooled
CCD videocamera (Photometrics Image Point, Pho-
tometrics, Tuscon, AZ).
of the immunohistochemical
staining.
Ila (1:1000) and Ki-Sl (1:
In Situ Hybridization
Frozen sections were cut at -200C and thawed onto
Superfrost Plus (Menzel, Germany) slides. Two syn-
thetic oligonucleotide probes directed against topo
Ila mRNA (nucleotides 2382 to 2427 and 3656 to
3701, GenBank accession number J0408834) were
labeled at the 3' end with [33P]dATP (DuPont-NEN
Research Products, Boston, MA) using terminal de-
oxynucleotidyltransferase (Amersham International,
Buckinghamshire, UK). A detailed description of the
hybridization method has been published
where.35 Briefly, unfixed, frozen, and air-dried sec-
tions were hybridized at 42°C for 18 hours with 1 x
107 cpm/ml of the probe, washed four times (15
minutes each) in 1 x standard saline citrate at 550C,
else-

Page 4

2076
4/PJune 1996, Vol. 148, No. 6
Jarvinen et al
and, while in the final rinse, left to cool at room
temperature (approximately 1 hour). Autoradiograph
films (Amersham ,3-max; Amersham International)
were overlaid on slides, exposed for 2 weeks, and
developed using LX24 developer and AL4 fixative
(Kodak, Rochester, NY). Histology was controlled
afterwards by staining the hybridized tissue sections
with hematoxylin.
Autoradiograph films were analyzed as described
previously elsewhere.36 Briefly, the measurements
were done by an image analysis system, and the
optical densities of radiographic films (representing
mRNA expression) were than converted and ex-
pressed as + for low expression, ++ for moderate
expression, and + + + for high expression of topo IIa
mRNA by one investigator (J. Kononen) without prior
knowledge of the immunohistochemical topo
data. Evaluation was done only on the histologically
most representative areas. The quantitation between
different hybridizations was based on the topo
mRNA levels of the breast cancer cell lines BT 474,
MCF-7, and UACC 812, which were included in each
hybridization experiment and formed an internal
standard for semi-quantitative analysis. The border-
line cases were scored as the lower of the two ex-
pression categories in question.
Ila
Ila
DNA Flow Cytometry
DNA flow cytometry was performed using 200-,um-
thick frozen sections as starting material.37 An EPICS
C flow cytometer (Coulter Electronics, Hialeah, FL)
was used for data acquisition, and a multicycle soft-
ware program (Phoenix Flow Systems, San Diego,
CA) was used for DNA histogram analysis. A DNA
index over 1.07 (in more than 20% of cells) was used
as a criterion for DNA aneuploidy. In DNA-aneuploid
cases, only the S phase of the aneuploid clone was
registered. Cell cycle evaluation was successful in
77% (177/230) of the tumors.
Statistical Methods
Statistical analyses were carried out with an IBM-
compatible personal computer and the Biomedical
Data Processing software (BMDP Statistical Soft-
ware, Los Angeles, CA). Association of topo Ila and
individual clinicopathological variables was done us-
ing one-way analysis of variance (ANOVA with Welch
correction for unequal variances). Adjustment of
topo Ia for the S-phase fraction was done by ana-
lyzing the ratio of topo Ila to the S-phase fraction.
Analysis of co-variance was used to evaluate the
proliferation-independent associations of topo Ila.
Results
ImmunohistochemicalStaining for Topo I1a.
Immunohistochemical stainings using different con-
centrations of topo Ila antibody showed that 1:1000
dilution of topo Ila polyclonal antibody yielded the
best result, a strong nuclear immunoreaction without
significant background staining (Figure 1, A and B).
Positive immunostaining was restricted to the nuclei
of carcinoma cells in all 230 invasive breast cancer
samples whereas stroma and nonmalignant breast
epithelium remained always negative. The propor-
tion of topo-Ila-immunopositivecells in breast carci-
nomas varied from 0.6 to 37.4% (10.6±7.9%, mean
±SD). The specificity of topo Ia was confirmed with
Western blots, which showed a single immunoreac-
tive band at 170 kd, which is in good agreement with
the molecular mass of human topo Ila32,33(Figure 2).
The specificity of topo Ila immunohistochemistry was
demonstrated by conventional immunohistochemi-
cal controls (not shown) and by a double-immunoflu-
orescence experiment in which two different topo
Ila-specific antibodies labeled exactly the same
cells (Figure 3, A and B).
In Situ Hybridization of Topo I/a.
Topo Ila gene expression was studied with mRNA in
situ hybridization using synthetic oligonucleotide
probes. Close association was obtained between the
topo Ila score by immunohistochemistry and for topo
Ia mRNA expression from adjacent tissue sections
(P = 0.007, n = 22, ANOVA; Figures 4 and 5). As a
control for specificity, the two distinct topo Ila oligo-
nucleotide probes gave identical hybridization sig-
nals when hybridizated separately on adjacent tis-
sue sections (data not shown). Addition of a 100-fold
excess of respective unlabeled probe into the hy-
bridization cocktail abolished the hybridization sig-
nal completely (not shown).
Association of Topo I/a Score with
Clinicopathological Features of the Tumors
Topo Ila expression was not associated with the
age of the patient, primary tumor size, or axillary
nodal status (P > 0.05; Table 1). Topo Ila scores
were significantly higher in ER- and PR-negative
tumors compared with positive ones (P < 0.0001
for both), and there was a strong association be-
tween high topo Ila scores and DNA aneuploidy (P
= 0.0003; Table 1). Poorly differentiated breast
carcinomas (grade
III) had significantly higher

Page 5

Topo Ila Expression in Breast Cancer
A/P June 1996, Vol. 148, No. 6
2077
Figure 1. Inniiiiinohistochemical staining of hbu1ian breac-st carcinionias with apoll'clonal topo IIa an2tibody. Tuniors witb a high (A) and lou, (B)
proportion ofjtopo-lHa-positii'e breast cancer cells are shown. Hemato:x1ylin counterstain; inagnification, X320.
topo
significantly higher topo
tumors (P < 0.0001, ANOVA; Table 1). The topo Ila
scores were high in tumors with c-erbB-2 oncopro-
tein overexpression (P < 0.0001). The proliferative
Ila scores than grade
11 tumors, which had
Ila scores than grade
activity of the tumors was determined by flow cy-
tometric S-phase analysis. In the linear regression
analysis, topo
cantly with S-phase fraction (r = 0.46, P < 0.0001;
Figure 6).
Ila score correlated highly signifi-