Overexpression of h-prune in Breast Cancer is Correlated with Advanced Disease Status
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
Vol. 11, 199 – 205, January 1, 2005 Clinical Cancer Research 199
Overexpression of h-prune in Breast Cancer is Correlated
with Advanced Disease Status
Massimo Zollo,1,6 Alessandra Andrè,1 Antonio Cossu,2 Conclusions: Although not significantly correlated with
Maria C. Sini,3 Anna D’Angelo,1 Natascia Marino,1 overall survival, positive h-prune immunostaining identifies
subsets of breast cancer patients with higher tumor
Mario Budroni,4 Francesco Tanda,2 aggressiveness. Further investigations using larger collec-
Gianluigi Arrigoni,5 and Giuseppe Palmieri3 tions of advanced breast cancer patients are required for
1
Telethon Institute of Genetics and Medicine, Naples, Italy; 2Istituto assessing the predictive role of h-prune in breast cancer.
Anatomia Patologica, Università di Sassari; 3Istituto di Chimica
Biomolecolare-Sezione di Sassari, Consiglio Nazionale delle Ricerche,
Località Tramariglio; 4Centro Multizonale di Osservazione INTRODUCTION
Epidemiologica, Azienda U.S.L.1, Sassari, Italy; 5Ospedale San Raffaele, Breast cancer is a complex disease, which has a difficult
HSR, Departimento di Patologia, Milan, Italy; and 6Centro di ricerca per clinical management due to its biological heterogeneity and its
l’ingegneria genetica di Napoli, Naples., Italy
wide spectrum of responsiveness to different treatments (1). Over
the past few years, knowledge of the molecular mechanisms
ABSTRACT underlying tumorigenesis have allowed the identification of an
Purpose: The h-prune gene is involved in cellular increasing number of biomarkers, which have been correlated
motility and metastasis formation in breast cancer through with cancer prognosis or used as predictors of response to specific
interacting with the nm23-H1 protein. The aim of this treatments, with varying degrees of success (1, 2). The well-
study was to better define the clinical and pathologic role established prognostic factors currently used in cases of primary
of h-prune in breast cancer patients. breast cancer include axillary lymph node involvement, histologic
Experimental Design: Using immunohistochemistry, we subtype, tumor size, nuclear or histologic grade, estrogen and
assessed h-prune and nm23-H1 protein expression in two progesterone receptor (ER and PR, respectively) status, and
series of breast cancer patients: (i) in 2,109 cases with proliferative index (3, 4). Novel tumor markers with potential
pathologic reports on primary tumors and (ii) in 412 cases clinical utility are now awaited, starting from newly identified
with detailed clinical information. To assess the role of DNA molecules that are involved in cell transformation, invasion, and
amplification in gene activation, the h-prune copy number metastasis.
was evaluated by fluorescence in situ hybridization analysis The human homologue of Drosophila prune (h-prune)
in 1,016 breast cancer cases. protein belongs to the DHH superfamily of phosphoesterases,
Results: In the patients tested (n = 2,463), 1,340 (54%) which have a cytoplasmic cyclic nucleotide phosphodiesterase
had an increased level of h-prune expression; a positive activity (5, 6). We reported previously that h-prune is involved
immunostaining for nm23-H1 was observed in 615 of 2,061 in both promoting cellular motility and stimulating expression
(30%) cases. Overexpression of h-prune was associated with of genes involved in metastatic pathways (7 – 9). In this
multiple gene copy number at chromosome 1q21.3 in a very respect, h-prune physically interacts with the nucleoside
limited fraction of cases (68 of 1,016; 6.7%), strongly diphosphate kinase nm23-H1, a known suppressor of cancer
indicating that alternative pathways induce h-prune activa- metastasis (7). Our recent findings indicate that h-prune may
tion in breast cancer. Multivariate Cox regression analysis have a role in the metastatic process through specific in-
showed that neither h-prune overexpression nor decreased hibition of the antimetastasis function of nm23-H1 ‘‘in vivo’’
nm23-H1 immunostaining is independent prognostic factors. (8, 9). Indeed, overexpression of h-prune seems to be involved
However, a significant association of h-prune overexpression in cancer progression and tumor aggressiveness (8, 9). One
with either advanced lymph node status (P = 0.017) or potential mechanism leading to h-prune activation may be
presence of distant metastases (P = 0.029) was observed. through the amplification of gene copy number, which has
been shown to induce cell proliferation and increased
expression levels of h-prune (8, 9). Thus, inhibition of h-prune
activity could interfere with the establishment of metastases
Received 7/22/04; revised 9/13/04; accepted 10/7/04. and represent a new target for future cancer therapy (9).
Grant support: Associazione Italiana Ricerca sul Cancro (M. Zollo In the present study, we examined the distribution of both
and G. Palmieri), Compagnia San Paolo Torino (M. Zollo), Ricerca
Finalizzata Ministero della Salute (A. Cossu and G. Palmieri), Regione h-prune and nm23-H1 expression in large and well-characterized
Autonoma della Sardegna (M. Budroni and F. Tanda), and Telethon cohorts of invasive breast carcinomas to determine their
Institute of Genetics and Medicine-Telethon Regione Campania (M. association with clinical and pathologic variables, as well as
Zollo). EU-FP6 BRECOSM, LSH-CT-503234 (M. Zollo). with patients’ prognosis.
The costs of publication of this article were defrayed in part by the
payment of page charges. This article must therefore be hereby marked
advertisement in accordance with 18 U.S.C. Section 1734 solely to
indicate this fact. PATIENTS AND METHODS
Requests for reprints: Massimo Zollo, Telethon Institute of Genetics
Patients. Patients were collected and considered eligible if
and Medicine, Via Pietro Castellino 111, 80131 Naples, Italy. Phone:
39-081-6132218; Fax: 39-081-6132351; E-mail: zollo@tigem.it. they had a histologic diagnosis of invasive breast carcinoma.
D2005 American Association for Cancer Research. Cases were retrieved from two archives. The first, from the200 H-prune and Clinicopathologic Studies
University Hospital of Basel (Switzerland), which included 1,531 of healthy breast tissue. In normal breast tissue, nm23-H1 was
invasive ductal carcinomas, 310 invasive lobular carcinomas, 69 homogeneously expressed, whereas expression of h-prune was
mucinous carcinomas, 65 tubular carcinomas, 48 medullary absent or of low intensity.
carcinomas, and 86 other types of invasive carcinomas. Formalin- Staining was evaluated semiquantitatively, using 54
fixed (4%; buffered), paraffin-embedded tumor samples were thus normal samples randomly positioned in duplicate across the
available from the Institute of Pathology, University Hospital multiple arrays. Intensity and distribution of immunohisto-
Basel; the Institute for Clinical Pathology, Basel; and the Triemli chemistry staining was used to classify tumor samples as
Hospital, Zurich. The second archive was from the Institute of positive (strong [+++] to moderate [++] staining, homoge-
Pathology at the University of Sassari (Italy) and included 307 neously distributed or presented by large majority of tumor
invasive ductal carcinomas, 69 invasive lobular carcinomas, 12 cells) or negative (absent or weak staining [+]) for both h-prune
mucinous carcinomas, and 24 other types of invasive carcinomas. and nm23 expression. For cases with discordant results between
All of the slides from all of the tumors were reviewed by at least the expert pathologists, agreement was achieved after discus-
three experienced pathologists who participated in the study sion at the microscope.
(A.C., G.A., and M.C.S.). Fluorescence In situ Hybridization Analysis. Paraffin-
All of the clinicopathologic features for each patient, embedded tissue microarray sections were treated according to
including disease stage at diagnosis, therapy, relapse, disease-free previously reported protocols (13, 14). The PAC 279-H19 clone,
survival, and time of last control (for overall survival), were either spanning the h-prune gene region at chromosome 1q21.3, and a
obtained from the cancer registry of Basel (10) and Sassari (11) or DNA/BAC clone specific for the nm23-H1 gene at chromosome
abstracted from the hospital records. The pathologic stage, the 17q21.3 were labeled by nick translation with dUTP-CY3 (red)
histologic grade (according to Elston and Ellis 12), and the nodal and used as probes. The pUC177 clone, corresponding to the
status were obtained from the primary pathology reports. pericentromeric region at 1q12, and the pZ17-14 clone,
Disease status at the time of diagnosis was defined corresponding to the centromeric region of chromosome 17,
depending on the clinical stage, as assessed by medical history, were labeled by nick translation with dCTP-FluorX (green) and
physical examination, blood cell count with white cell differential used as controls. The pUC177 clone was kindly provided by Dr.
and biochemistry, and instrumental tests (chest radiography, Mariano Rocchi (Istituto di Genetica, Universitá degli Studi di
liver CT scan and/or ultrasound, bone nuclear scan, etc.). Up to Bari, Bari, Italy). Nuclei were counterstained with 4V,6-
10-year clinical follow-up data were available for 2,299 cases diamidino-2-phenyl-indole. Two distinct experiments were done
(median = 56 months, range = 2 – 120 months); 15-year clinical for each case. Digital images were captured using an Olympus
follow-up data were available for 222 patients (median = 138 BX-61 epifluorescence microscope equipped with the appropri-
months, range = 121-176 months). ate filters for excitation of 4V,6-diamidino-2-phenyl-indole, Cy3
The use of the specimens and data for this study was (orange) or FluorX (vysis), and with a COHU video and
approved by the Ethics Committees at both Basel University Cytovision software. Hybridization signals on at least 100 intact,
Hospital and the University of Sassari. well-preserved, and nonoverlapping, nuclei were evaluated by at
Tissue Microarrays. Tissue microarray construction was least two investigators.
as previously described (10, 13). Briefly, tissue cylinders with a Statistical Analyses. The following variables and cate-
diameter of 0.6 mm were punched from representative tumor areas gories were defined and included in our analyses: pathologic
of a ‘‘donor’’ tissue block using a home-made semiautomatic primary tumor size, pathologic nodal status, presence of
robotic precision instrument. These were positioned in six metastases, ER and PR status, age at diagnosis, histologic
different recipient paraffin blocks, each of which contained tumor type, and overall survival (calculated starting from the
between 342 and 522 individual samples. Four-micrometer time of diagnosis). Some of these variables were missing for
sections of the resulting tissue microarray blocks were transferred some of the patients. In particular, nodal status, presence of
to an adhesive coated slide system (Instrumedics, Inc., Hack- metastases, and receptor status were not known in a large
ensack, NJ) following standardized procedures (10, 13). The fraction of the cases because these variables were not required
presence of tumor tissue on the multiple-arrayed sample was for the inclusion of the patients in the study.
verified by H&E staining. Tissue microarrays included at least two The survival data analysis was carried out with the
core sections from different areas of breast cancer from each statistical package Egret (version 2.0.3). The Cox regression
patient as well as several normal tissues as controls. model was done using raw mortality and tumor-specific
Immunohistochemistry. Sections from formalin-fixed, mortality. The time of overall survival was expressed in months,
paraffin-embedded tissues were immunostained using an anti-h- and the independent variables (h-prune, nm23-H1, pathologic
prune monoclonal antibody (clone 4G3/4, raised against a primary tumor size, and pathologic nodal status) were stratified
recombinant fusion protein of amino acids 1 – 351; Apotech Co., in three age groups: 23 to 44 (n = 252), 45 to 64 (n = 1,146), and
Switzerland,) and an anti-nm23-H1 antibody (clone K73, specific 65 to 98 (n = 1,123). Kaplan-Meier estimates were executed
for the H1 isoform; Apotech). The specificity of both antibodies through stratifying by h-prune and nm23-H1 immunostaining
was previously characterized in detail (8, 9). Immunohistochem- data. The Pearson’s v 2 test was used for assessing h-prune and
istry analysis was done using the Vectastain Elite ABC Kit (Vector nm23-H1 expression, with the abovementioned pathologic
Laboratories, Inc., Burlingame, CA) according to the manufac- variables (histologic tumor type, pathologic primary tumor size,
turer’s instructions and following a previously described protocol pathologic nodal status, presence of metastases, ER, and PR).
(8, 9). Optimized immunohistochemistry protocols were estab- The exact coefficient for sample proportion analysis was
lished by staining representative control histopathology sections calculated to determine all of the significant variables (Clinical Cancer Research 201
level). All analyses were done with the statistical package SPSS/ a positive h-prune immunostaining in tumors with at least
7.5 for Windows. trisomy (P = 0.027) or tetrasomy (P = 0.033; Table 1B).
However, the low frequency of such a cytogenetic alteration
strongly suggests that alternative pathogenetic pathways are
RESULTS involved in determining h-prune activation and the increased
Immunohistochemical Analysis. Assessment of h-prune somatic expression of h-prune in breast cancer.
and nm23-H1 expression levels, as well as of h-prune Correlation between h-prune and Clinicopathologic
chromosomal copy number, was carried out on tissue microarray Variables. To evaluate the clinicopathologic role of h-prune
sections from the archival tissues of 2,109 patients with both a overexpression in breast cancer, immunohistochemistry analy-
histologically proven diagnosis of breast carcinoma and ses were done on invasive primary breast tumors using both
available follow-up data. Additional tissue microarray sections the series of 2,109 breast cancer cases (with all of the
from 412 breast cancer patients with a well-assessed stage of the pathologic primary tumor information, and up to 15-year
disease (i.e., whose tumor-node-metastasis classification was clinical follow-up data; Table 2A) and the series of 412 breast
available) were evaluated for h-prune immunostaining. Overall, cancer cases (with a complete tumor-node-metastasis (TNM),
the majority of patients had ductal carcinoma as the histologic system for staging cancer classification, and up to 10-year
variant (1,838; 73%) and were >60 years of age (1,425; 57%) at clinical follow-up data; Table 3). Tumor sections from the
the time of diagnosis. Records of the clinical follow-ups for each subset of 2,109 patients were also investigated for nm23-H1
patient were available, covering a median period of 59 months expression by immunohistochemistry analysis (Table 2B). This
(range = 2 – 176); the majority of the patients were still alive latter subset was uninformative for the presence of distant
(1,716; 68%), and only a very few cases were lost in the follow- metastases; information on nodal status was partially available
up (12; 0.5%) at the time of this study. for such breast cancer patients in terms of presence (N+) or
Expression of h-prune and nm23-H1 was evaluated by absence (N) of lymph node involvement (positive h-prune
immunohistochemistry analysis using two specific antibodies immunostaining was observed in 210 of 482 [44%] N+ cases
(4G3/4 and K73, respectively). Figure 1A and B shows
representative examples of immunohistochemistry staining for
h-prune in cases from a tissue microarray series. Altogether, a
strongly positive cytoplasmic immunostaining for h-prune
(referred to as h-prune +) was observed in the majority (1,340;
54%) of the 2,463 tumors tested (58 cases were not assessable);
conversely, a positive cytoplasmic immunostaining for nm23-H1
was observed in 615 (30%) of the 2,061 tumors tested. Among the
2,061 tumor tissues evaluated for both h-prune and nm23-H1
expression, an inverse distribution of positive immunostaining
was observed (1,180 [57%] breast carcinomas were h-prune+,
whereas 615 [30%] were nm23-H1+; Table 1A). No statistical
correlation between h-prune and nm23-H1 expression was
observed (Table 1A).
Fluorescence In situ Hybridization Analysis. Fluores-
cence in situ hybridization (FISH) analysis was done on 1,016
assessable tumors from breast cancer patients using as probes a
PAC clone (279-H19) corresponding to the h-prune genomic
region at chromosome 1q21.3, and a control clone spanning the
pericentromeric region at 1q12. Multiple FISH signals at 1q21.3
in >20% of the analyzed nuclei were found in 173 (17%) cases
(Table 1B; Fig. 1C). Considering the tumors with at least two
gene copies per centromere (i.e., at least tetrasomic signals) and
even including breast carcinomas with polysomy of entire
chromosome 1 due to the simultaneous presence of multiple
centromere signals, a very low level of h-prune amplification
at chromosome 1q21.3 was observed (68 of 1,016; 6.7%;
Table 1B). Using the BAC clone specific for the nm23-H1 gene
at chromosome 17q21.3 and the pZ17-14 clone at centromeric
region of chromosome 17 as control, no cytogenetic alteration
was found at this level in our series. Absence of karyotypic
anomalies in cells from normal tissues surrounding the tumors
strongly indicated that amplification of the h-prune genomic Fig. 1 Expression and cytogenetic analysis of h-prune in breast
carcinomas. A, IHC analysis of h-prune expression in representative
region is highly specific for breast cancer cells. TMA sections; 40x magnification of positive (a) and negative (b) h-prune
The increase in DNA copy number at the h-prune immunostaining. B, FISH analysis on the same samples using h-prune/
genomic region was significantly associated to the presence of PAC279-H19 (red) and control pUC177 (green) as probes.202 H-prune and Clinicopathologic Studies
Table 1 Results of immunohistochemistry and FISH analyses for h-prune in the 2,109 series of invasive breast carcinomas
A. Distribution of h-prune and nm23-H1 immunohistochemistry staining
Immunohistochemistry staining nm23-H1 Total P
Positive Negative
h-prune positive 359 (30%) 821 (70%) 1,180 0.219
negative 256 (29%) 625 (71%) 881
B. Correlation between immunohistochemistry and FISH results for h-prune
Marker n Disomy (%) Trisomy or more* (%) P Tetrasomy or morey (%) P
h-prune negative 440 386 (88) 54 (12) 0.027 21 (4.8) 0.033
h-prune positive 576 457 (79) 119 (21) 47 (8.2)
Total tested 1,016 843 (83) 173 (17) 68 (76.7)
NOTE: n, number of cases; P, v 2 Pearson’s test; two tailed; 95% CI.
*Multiple signals in >20% of the analyzed nuclei after hybridization with a PAC clone corresponding to the h-prune gene at chromosome 1q21.3.
yIncluding also amplification of the entire chromosome 1.
and 179/496 [36%] N cases, P = 0.109; analogously, no As of December 2002, 782 (31%) patients have died due
statistical correlation for nm23-H1+ immunostaining was to disease, with the median overall survival of the whole
found in these two groups). Due to this fragmentary sample being 59 months, with a median follow-up of live
information on progression-free survival in our series, the patients of 72 months. Using Pearson’s v 2 test, h-prune and
prognostic values of each variable were determined on the nm23-H1 immunostaining were evaluated for association with
basis of the overall survival analysis. several pathologic variables: histologic type, tumor grading
Table 2 Correlation between IHC results and pathologic variables in the series of 2,109 breast cancer patients
A. Correlation of h-prune with tumor features
Tumor variable h-prune h-prune Total P
Negative Positive Negative Positive
Size T1 319 (42%) 440 (58%) T1-2 765 (44%) 985 (56%) 1,750 0.63
T2 446 (45%) 545 (55%)
T3 54 (45%) 65 (55%) T3-4 160 (45%) 196 (55%) 356
T4 106 (45%) 131 (55%)
Grading G1 211 (40%) 323 (60%) G1-2 584 (41%) 831 (59%) 1,415 0.13
G2 373 (42%) 508 (58%)
G3 336 (48%) 357 (52%) G3 336 (48%) 357 (52%) 693
Histology Ductal 621 (42%) 865 (58%) 1,486 0.42
Lobular 117 (40%) 174 (60%) 291
Other 125 (49%) 131 (51%) 256
ER Positive 649 (42%) 882 (58%) 1,531 0.37
Negative 204 (44%) 259 (56%) 463
PR Positive 279 (43%) 369 (57%) 648 0.34
Negative 523 (42%) 714 (58%) 1,237
B. Correlation of nm23-H1 with tumor features
Tumor variable nm23-H1 nm23-H1 Total P
Negative Positive Negative Positive
Size T1 489 (68%) 226 (32%) T1-2 1,162 (69%) 517 (31%) 1,679 0.27
T2 673 (70%) 291 (30%)
T3 86 (74%) 31 (26%) T3-4 254 (73%) 94 (27%) 348
T4 168 (73%) 63 (27%)
Grading G1 360 (69%) 159 (31%) G1-2 937 (68%) 434 (32%) 1,371 0.14
G2 577 (68%) 275 (32%)
G3 501 (74%) 179 (26%) G3 501 (74%) 179 (26%) 680
Histology Ductal 1,031 (69%) 455 (31%) 1,486 0.68
Lobular 201 (69%) 90 (31%) 291
Other 167 (65%) 89 (35%) 256
ER Positive 1,076 (71%) 447 (29%) 1,523 0.41
Negative 329 (70%) 142 (30%) 471
PR Positive 449 (69%) 203 (31%) 652 0.27
Negative 871 (71%) 362 (29%) 1,233
NOTE: P, v 2 Pearson’s test; two tailed; 95% CI.
Abbreviations: ER, estrogen receptor status; PR, progesterone receptor status.Clinical Cancer Research 203
Table 3 Correlation between h-prune expression and pathological variables in the series of 412 breast cancer patients
TNM h-prune h-prune Total P
Negative Positive Negative Positive
Tumor size T1 83 (58%) 59 (42%) T1-2 170 (58%) 122 (42%) 292 0.099
T2 87 (58%) 63 (42%)
T3 11 (44%) 14 (56%) T3-4 26 (47%) 29 (53%) 55
T4 15 (50%) 15 (50%)
Nodal status N0 125 (71%) 50 (29%) N0-1 230 (68%) 109 (32%) 339 0.017
N1 105 (64%) 59 (36%)
N2 6 (30%) 14 (70%) N2-3 6 (27%) 16 (73%) 22
N3 — 2 (100%)
Metastasis M0 210 (67%) 103 (33%) 313 0.029
M1 14 (36%) 25 (64%) 39
NOTE: P, v 2 Pearson’s test; two tailed; 95% CI.
Abbreviations: TNM, tumor-node-metastasis.
(as standardized by Elston and Ellis), pathologic primary tumor and the resultant decrease in the amount of free, unbound
size, ER, PR, and (when available) pathologic nodal status and cellular nm23-H1 molecules (8, 9). This interaction with nm23-
presence of metastases. No statistically significant correlations H1 greatly increases the h-prune phosphodiesterase activity,
between either h-prune or nm23-H1 expression and tumor type and when this is associated with high levels of h-prune, it
(ductal versus lobular), histologic grade, pathologic primary induces positive effects on cellular motility both in vitro and
tumor size, ER, and PR reactivity were observed (Table 2). in vivo (8, 9). These effects have been shown to promote
Positive h-prune immunostaining was instead significantly metastasis formation in breast cancer, and seem to be reduced
associated with either the advanced nodal status (N2 – N3 by an anti-h-prune phosphodiesterase inhibitor (9, 15, 16).
group; P = 0.017) or the presence of distant metastases (M1 Overexpression and increased enzymatic activity of h-prune
group; P = 0.029) among the series of 412 breast cancer has thus been suggested to be involved in promoting cancer
patients (Table 3). Statistical analysis of this series confirmed metastasis through the alteration of nm23-H1 metastasis
the absence of any significant association between h-prune suppressor activity (17). In particular, it has been suggested that
overexpression and primary tumor size (Table 3).Using the Cox cellular motility may be strongly induced by the amplification
model adjusted according to disease stage and age at diagnosis, of h-prune in tumor cells in vitro, and that mobilization of
no prognostic values of either h-prune overexpression [hazard neoplastic cells from the site of the primary lesion is necessary,
ratio, 0.61; 95% confidence interval (95% CI), 0.31 – 1.18; P = although not sufficient, to produce distant metastases (8, 9). Our
0.144] or nm23-H1 immunostaining (risk ratio, 0.98; 95% CI, findings through FISH analysis have shown that increase of the
0.90 – 1.07; P = 0.158) were observed (Table 4; Fig. 2). h-prune copy number is an infrequent event in breast cancer.
Overall, tumor grading, primary tumor size, and axillary nodal Such a discrepancy with previous FISH data indicating a higher
status always remained the variables that were closely frequency of h-prune amplification in breast cancer (although
correlated to prognosis in this series of breast cancer patients these results have been obtained in series with a much more
(Table 4). limited number of cases; refs. 8, 9) suggests that different
pathogenetic mechanisms may concur for h-prune activation and
protein overexpression during breast cancer progression. In this
DISCUSSION
respect, similar observations have been inferred by comparing
In this study, we have shown that neither high expression
data from CGH analysis and whole-gene expression in several
levels of h-prune, which is a member of the DHH superfamily of
breast cell lines (18 – 20). Indeed, a significant number of new
phosphoesterases, or down-regulated expression of nm23-H1,
which physically interacts with h-prune, have a role as
independent predictive prognostic factors in the clinical outcome Table 4 Correlation of histopathological and IHC markers with overall
for invasive breast carcinomas. However, h-prune overexpres- survival
sion seems to be associated with the presence of either advanced
A. Statistical analysis of the series of 2,109 breast cancer patients
nodal status (N2 – N3 cases) or distant metastases (M1 cases),
strongly suggesting that it can be used as a marker for the Marker Risk ratio 95% CI P
identification of subsets of breast cancer patients with a more h-prune (negative versus positive) 0.93 0.78-1.09 0.369
aggressive disease. nm23-H1 (negative versus positive) 0.98 0.90-1.07 0.158
Tumor size (pT1 versus pT2-4) 0.81 0.68-0.96 0.017
The h-prune protein has been previously shown to be Tumor grading (G1 versus G2-3) 0.70 0.59-0.83 0.001
expressed at high levels in human sarcomas and breast
B. Statistical analysis of the series of 412 breast cancer patients
carcinomas (8). For h-prune function in breast cancer cells,
we previously showed the relevance of its cyclic nucleotide Marker Hazard ratio 95% CI P
phosphodiesterase activity in its inhibition of nm23-H1, the h-prune (negative versus positive) 0.61 0.31-1.18 0.144
gene for which is known to act as a suppressor of metastasis. Tumor size (pT1-4) 12.49 4.10-38.02204 H-prune and Clinicopathologic Studies
gene expression that affects cancer progression (a 2-fold change
in DNA copy number has been definitively associated with a
corresponding 1.5-fold change in mRNA levels; refs. 18 – 20).
On the other hand, there is no clear explanation for the
apparently non-specific presence of increased gene copy number
in cases with normal expression of h-prune (although, near-
diploid karyotypes with few or single structural or numerical
anomalies have been reported to be a specific feature in the
initial stages of neoplasia; refs. 21, 22).
In the present study, we have analyzed a large series of breast
carcinomas and found somatic overexpression of h-prune in the
majority of cases tested (1,340 of 2,463; 54%), as determined by
immunohistochemistry analysis with tissue microarrays (the
advantage of using the tissue microarray approach is the improved
standardization, capacity, and speed of the analyses; refs. 10, 13).
Our data indicate that strong h-prune immunoreactivity is highly
specific for breast cancer cells, since the normal counterpart
(epithelial cells) in each case of breast carcinoma consistently
failed to express h-prune at comparably high levels (see Fig. 1).
However, h-prune+ cases were more or less equally distributed
among the different patients’ subsets, and no correlations with
histologic tumor type, primary tumor size and grading, ER and PR
status were observed. Conversely, a significant association of h-
prune overexpression with either advanced axillary nodal status
or presence of distant metastases was found in our series.
Multivariate analyses revealed that both h-prune and nm23-H1
protein levels were not correlated with the overall survival in
breast cancer patients, thus not adding precision to the predictive
power of the stage of disease (see Tables 2 – 4). Overall, one could
speculate that standardized adjuvant therapies or other, as-yet-
unidentified, factors may affect the predictive role of the
alterations within the h-prune/nm23-H1 molecular pathway.
Overexpression of h-prune has been shown to interfere with
the nm23-H1 antimetastatic activity through a physical protein-
protein interaction (7). Increased levels of nm23-H1 expression
have been shown to be associated with early disease stages,
whereas a lack of nm23-H1 expression has been associated with
more advanced stages and more aggressive tumors (8, 9).
Conversely, our findings have shown the absence of any
correlation between nm23-H1 somatic expression and extent of
disease, in terms of primary tumor size and nodal status.
As the presence or absence of distant metastases is a
determining factor in the clinical course of breast cancer, the lack
of an association between loss of nm23-H1 expression and poor
prognosis strongly argues against any association between
nm23-H1 expression and distant metastases in our experimental
system. Although distal dissemination has been associated with
nm23-H1 levels in primary tumors (23), other studies have also
failed to identify any prognostic significance of nm23-H1
Fig. 2 Kaplan_Meier cumulative survival analyses according to h- expression in breast cancer patients (24, 25). Data on distant
prune and nm23-H1 status. All cases of breast carcinomas evaluated
tumor dissemination are lacking in patients analyzed for nm23-
for positive immunostaining of nm23-H1 (A), and h-prune in the series
of 2,109 patients (B), or h-prune in the series of 412 patients (C). In A, H1 expression from our series, and the establishment of distant
(dashed line) nm23 positive versus (continous line) nm23 negative. In metastases has been shown to be particularly associated with the
B, (dashed line) h-prune negative, (continous line) h-prune positive. In levels of nm23-H1 expression (breast cancer patients with
C, (dashed line) h-prune negative, (continous line) h-prune positive. lesions positive for nm23-H1 seem to present significantly
reduced distant recurrence-free survival; refs. 23 – 25). Overall,
potential breast markers are located in chromosomal regions with the real impact of nm23-H1 expression levels on clinical
low levels of DNA gain (2-4 copies per genome) and are outcome is still a controversial issue in human cancers
sufficient to actively participate in the global deregulation of (23 – 28) and needs further investigation. On this regard, weClinical Cancer Research 205
started collecting a larger number of additional breast cancer 11. Cesaraccio R, Sechi O, Pirino DR, et al. Incidenza dei tumori negli
cases with more detailed clinicopathologic information, in order anni 1993 – 97 in Provincia di Sassari. In: Budroni M, Tanda F, editors.
Registro Tumori della Provincia di Sassari. I tumori in Sardegna negli
to increase the number of N2-3 and M1 groups, therefore anni novanta. Perfugas, Sassari (Italy): Tipografia AM GRAPHIC; 2002.
obtaining a clear and definitive result about the role of the nm23- p. 9 – 22.
H1 pathway in such a disease. 12. Elston CW, Ellis IO. Pathological prognostic factors in breast cancer.
Identification of new surrogate markers as potential The value of histological grade in breast cancer: experience from a large
predictive factors will be useful to improve the clinical study with long-term follow-up. Histopathology 1991;19:403 – 10.
management of breast cancer patients (i.e., to identify which 13. Simon R, Sauter G. Tissue microarrays for miniaturized high-
throughput molecular profiling of tumors. Exp Hematol 2002;30:
subset of patients is expected to show either a response or a lack 1365 – 72.
of response to a particular therapy). Nevertheless, the prediction 14. Muresu R, Sini MC, Cossu A, et al. Chromosomal abnormalities and
of metastatic potential has been longer indicated as one of the microsatellite instability in sporadic endometrial cancer. Eur J Cancer
main goals to be pursued to assess the risk subgroups in patients 2002;38:1802 – 9.
with primary breast cancer (29, 30). Metastasis formation in 15. Haaz MC, Fischel JL, Formento P, Renee N, Etienne MC, Milano G.
breast cancer is now continually under investigation, and basic Impact of different fluorouracil biochemical modulators on cellular
research efforts are under way that are focused on the multistep dihydropyrimidine dehydrogenase. Cancer Chemother Pharmacol 1996;
38:52 – 8.
events occurring in tumor dissemination (i.e., motility, invasion,
16. Hejna M, Raderer M, Zielinski CC. Inhibition of metastases by
angiogenesis, and metastatic colonization), in which h-prune anticoagulants. J Natl Cancer Inst 1999;91:22 – 36.
seems to participate by increasing metastatic potential (9, 31). 17. Hartsough MT, Clare SE, Mair M, et al. Elevation of breast
Based on our findings, h-prune represents a new cell-type- carcinoma Nm23-H1 metastasis suppressor gene expression and
specific gene that is associated with disease progression in breast reduced motility by DNA methylation inhibition. Cancer Res 2001;
cancer patients. Further studies that must include cohorts of 61:2320 – 7.
patients undergoing different therapeutic approaches are required 18. Forozan F, Mahlamaki EH, Monni O, et al. Comparative genomic
hybridization analysis of 38 breast cancer cell lines: a basis for in-
to better understand the real clinical significance of the h-prune/ terpreting complementary DNA microarray data. Cancer Res 2000;60:
nm23-H1 pathway. 4519 – 25.
19. Pollack JR, Sørlie T, Perou CM, et al. Microarray analysis reveals
ACKNOWLEDGMENTS a major direct role of DNA copy number alteration in the transcriptional
We thank Drs. G. Sauter, M. Mirlacher, R. Simon from the Institute program of human breast tumors. Proc Natl Acad Sci U S A 2002;99:
of Pathology at the University Hospital of Basel (Switzerland), for 12963 – 8.
providing part of the tumor cases in this study and for helpful discussion. 20. Hyman E, Kauraniemi P, Hautaniemi S, et al. Impact of DNA
amplification on gene expression patterns in breast cancer. Cancer Res
2002;62:6240 – 5.
REFERENCES 21. Lengauer C, Kinzier KW, Volgestein B. Genetic instabilities in
1. Keen JC, Davidson NE. The biology of breast carcinoma. Cancer human cancers. Nature 1998;396:643 – 9.
2003;97:825 – 33. 22. Jasin M. Chromosome breaks and genomic instability. Cancer Invest
2. Domchek SM, Weber BL. Recent advances in breast cancer biology. 2000;18:78 – 86.
Curr Opin Oncol 2002;14:589 – 93. 23. Mao H, Liu H, Fu X, Fang Z, Abrams J, Worsham MJ. Loss of nm23
3. Kuru B, Camlibel M, Dinc S, Gulcelik MA, Alagol H. Prognostic expression predicts distal metastases and poorer survival for breast
factors affecting survival and disease-free survival in lymph-node- cancer. Int J Oncol 2001;18:587 – 91.
negative breast carcinomas. J Surg Oncol 2003;83:167 – 72. 24. Charpin C, Garcia S, Bonnier P, et al. Prognostic significance of
4. Morabito A, Magnani E, Gion M, et al. Prognostic and predictive Nm23/NDPK expression in breast carcinoma, assessed on 10-year
indicators in operable breast cancer. Clin Breast Cancer 2003;3:381 – 90. follow-up by automated and quantitative immunocytochemical assays.
J Pathol 1998;184:401 – 7.
5. Timmons L, Shearn A. Germline transformation using a prune cDNA
rescues prune/killer of prune lethality and the prune eye color phenotype 25. Gohring UJ, Eustermann I, Becker M, Neuhaus W, Rein DT,
in Drosophila. Genetics 1996;144:1589 – 600. Schondorf T. Lack of prognostic significance of nm23 expression in
6. Aravind L, Koonin EV. A novel family of predicted phosphoesterases human primary breast cancer. Oncol Rep 2002;9:1205 – 8.
includes Drosophila prune protein and bacterial RecJ exonuclease. 26. Hartsough MT, Steeg PS. Nm23/nucleoside diphosphate kinase in
Trends Biochem Sci 1998;23:17 – 9. human cancers. J Bioenerg Biomembr 2000;32:301 – 8.
7. Reymond A, Volorio S, Merla G, et al. Evidence for interaction 27. Niitsu N, Okabe-Kado J, Okamoto M, et al. Serum nm23-H1 protein
between human PRUNE and nm23-H1 NDP kinase. Oncogene 1999; as a prognostic factor in aggressive non-Hodgkin lymphoma. Blood
18:7244 – 52. 2001;97:1202 – 10.
8. Forus A, D’Angelo A, Henriksen J, et al. Amplification and 28. Nesi G, Palli D, Pernice LM, et al. Expression of nm23 gene in
overexpression of PRUNE in human sarcomas and breast carcinomas: gastric cancer is associated with a poor 5-year survival. Anticancer Res
a possible mechanism for altering the nm23-H1 activity. Oncogene 2001; 2001;21:3643 – 9.
20:6881 – 90. 29. Heimann R, Hellman S. Individual characterisation of the metastatic
9. D’Angelo A, Garzia L, Andre A, et al. Prune cAMP phosphodies- capacity of human breast carcinoma. Eur J Cancer 2000;36:1631 – 9.
terase promotes cancer metastasis by down-regulation of nm23-H1. 30. Clark GM. Interpreting and integrating risk factors for patients with
Cancer Cell 2004;5:137 – 49. primary breast cancer. J Natl Cancer Inst Monogr 2001;30:17 – 21.
10. Simon R, Panussis S, Maurer R, et al. KIT (CD117)-positive breast 31. D’Angelo A, Zollo M. Unraveling genes and pathways influenced
cancers are infrequent and lack KIT gene mutations. Clin Cancer Res by h-prune PDE overexpression: a model to study cellular motility. Cell
2004;10:178 – 83. Cycle 2004;3:758 – 61.You can also read
