Early Detection and Diagnosis
Prognostic significance of cyclin A in gastric cancer†
Article first published online: 17 MAY 2006
Copyright © 2006 Wiley-Liss, Inc.
International Journal of Cancer
Volume 119, Issue 8, pages 1897–1901, 15 October 2006
How to Cite
Mrena, J., Wiksten, J.-P., Kokkola, A., Nordling, S., Haglund, C. and Ristimäki, A. (2006), Prognostic significance of cyclin A in gastric cancer. Int. J. Cancer, 119: 1897–1901. doi: 10.1002/ijc.21944
There are no relationships resulting in an actual, potential or apparent conflict of interest with regard to this manuscript.
- Issue published online: 11 AUG 2006
- Article first published online: 17 MAY 2006
- Manuscript Accepted: 6 FEB 2006
- Manuscript Received: 15 SEP 2005
- Finnish Cancer Society
- Finska Läkaresällskapet
- Medicinska Understödsföreningen Liv och Hälsa
- Helsinki University Central Hospital
- Helsinki University, Academy of Finland
- Sigrid Jusélius Foundation
- cyclin A;
- gastric cancer;
High level of cyclin A promotes carcinogenesis, and overexpression of cyclin A has been associated with poor prognosis of cancer patients. We validated the prognostic role of cyclin A in gastric cancer and evaluated its correlation with expression of an mRNA stability factor HuR. From 342 consecutive histologically confirmed gastric cancer patients were obtained 325 representative tissue specimens for cyclin A and 316 for HuR immunohistochemistry. Specimens were stained by cyclin A and HuR specific monoclonal antibodies. Nuclear immunostaining detected in ≥≥≥≥5% of the tumor cells was considered the cut-off for cyclin A positivity. Positive HuR immunoreactivity was scored as nuclear or cytoplasmic. Associations between scores, clinicopathological factors and survival were calculated by the χ2-test, Fisher's exact test, Kaplan-Meier test and Cox model. Cyclin A detected in the nuclei of cancer cells was positive in 55% (179 of 325) of the specimens; 40% (127 of 316) of the specimens had cytoplasmic and 88% (279of 316) nuclear immunoreactivity of HuR. Cyclin A expression was an independent prognostic factor for poor survival. Cyclin A immunoreactivity was associated with old age, high stage, proximal location of the tumor, intestinal type, noncurative resection, advanced penetration depth and with nodal metastases but not distant metastases. Furthermore, cyclin A expression was associated with cytoplasmic HuR expression, whereas no association with nuclear HuR was evident. Cyclin A is an independent prognostic factor in gastric cancer, and one mechanism for its overexpression may depend on cytoplasmic localization of HuR. © 2006 Wiley-Liss, Inc.
Cyclin A belongs to the cyclin protein superfamily, and it can activate two different cyclin-dependent kinases, CDK1 and CDK2. The level of cyclin A accumulates progressively throughout the interphase and disappears rapidly at the end of mitosis. Currently, phosphorylated cyclin A-CDK complex is suggested to play an important role in the initiation of DNA replication in the Sphase. The precise function of cyclin A in mitosis is unclear, but it may prevent other cyclins from degradation. Overexpression of cyclin A and dysregulation of CDK–cyclin complexes promote tumor cell growth, which can be facilitated by phosphorylation of oncoproteins and tumor suppressors.1, 2
In gastric cancer, cell-cycle regulators have been connected with poor prognosis. For example, overexpression of cyclin E, which also binds to CDK2 and controls the G1-S transition, has been associated with aggressive disease.3, 4 Furthermore, strongly positive cyclin E in the tumors of patients having surgery with intent to cure is a marker of poor prognosis, and concurrent expression of cyclin E and p53 is an independent prognostic factor.5 However, there are also reports indicating that cyclin E expression does not correlate with clinicopathological parameters,6 or is associated with good prognosis.7 Overexpression of cyclin A correlates with poor outcome in various cancer patients2 but thus far has not been shown to affect prognosis of gastric cancer patients.8
During the cell cycle, expression of cyclins and cell cycle-related proteins is strictly regulated. Transcriptional regulation of cyclins is cell-cycle dependent,9 and posttranscriptional control includes rapid and specific proteolysis by the ubiquitin–proteasome pathway.10 It has been postulated that mRNA stability plays an important role in the expression of certain cyclin proteins. HuR is an mRNA-binding and stabilizing factor and is a member of the embryonic lethal abnormal vision (ELAV)/Hu protein family.11 Several genes encoding proto-oncogenes, cytokines and cytokine response genes contain unstable AU-rich elements (AREs). During translocation from nucleus to cytoplasm, HuR is bound to these labile transcripts and prevents them from degrading, providing efficient translation of the target protein.12 Colorectal carcinoma cells expressing reduced levels of HuR show reduction in cyclin A and cyclin B1 mRNAs, decreased expression of both proteins and reduced activity of the CDKs regulated by these cyclins, as well as reduced cell growth.13 These data suggest that HuR controls cyclin A expression and regulates cell cycle control and proliferation. Previously, expression of HuR has been associated with the poor prognosis of breast,14 ovarian15 and gastric cancers.16
The prognostic role of cyclin A in gastric cancer is unclear. In this study, we evaluated the association of cyclin A expression with clinicopathological parameters and survival in patients with gastric adenocarcinomas. We also investigated whether an association exists in these tumors between expression of cyclin A and HuR.
Material and methods
The patient material comprised 342 consecutive cases operated on for gastric adenocarcinoma at the Department of Surgery, Meilahti Hospital and Helsinki University Central Hospital, between 1983 and 1999. Approval for the study was obtained from the local ethics committee. Diagnosis was confirmed from histological samples, and staging was performed according to UICC classification (TNM 1992). Four of these patients were excluded because of incomplete information: missing clinical data (one), unclear cause of death (one), and missing follow-up (two). Of the remaining cases, 325 for cyclin A and 316 for HuR could be analyzed by immunohistochemistry. In 309 patients, both cyclin A and HuR stainings were successful. Median age was 66 years. All stages were represented, the majority of which were stages I to IIIA (197 patients, 61%). Surgery for cure (R0 resection) was performed in 165 cases (51%) and none received neoadjuvant therapy. Median follow-up time was 12.7 years (range, 4.7–20.8). Clinicopathological characteristics of the series are described in Table I. Survival data were obtained until February 2004 from the Finnish Cancer Registry and the Population Register Centre of Finland.
|Clinicopathological variable||N||High expression of Cyclin A, n (%)||p value|
|Stage IA||49||15 (31)|
|Stage IB||42||21 (50)|
|Stage II||39||23 (59)|
|Stage IIIA||67||43 (64)|
|Stage IIIB||29||20 (69)|
|Stage IV||99||57 (58)|
|Penetration depth (pT)||<0.001|
|Lymph node metastases||0.005|
|Proximal 1/3||65||45 (69)|
|Distal 2/3||233||122 (52)|
|Tumor size (cm) median||0.74|
|Intent to cure||165||79 (48)|
Formalin-fixed tumor samples were embedded in paraffin. Archive tissue blocks were freshly cut into 4-μm sections onto slides and dried to fix for 12–24 h at 37°C. Tissue sections were then deparaffinized in xylene and rehydrated through graded alcohol and deionized water. For antigen retrieval, the sections were heated in a microwave oven in 0.3% citrate buffer (pH 6.0) for 4× 5 min, and washed with phosphate-buffered saline (PBS) for 2 × 5 min. Endogenous peroxidases were inactivated by 30-min incubation in methanol containing 1.6% hydrogen-peroxidase, followed by a PBS wash for 3 × 5 min. To block nonspecific binding sites, the sections were treated with normal horse serum for 15min. The sections were incubated overnight with an antihuman cyclin A mouse monoclonal antibody corresponding to the N-terminal fragment of cyclin A protein (6E6, Novocastra Laboratories, Newcastle upon Tyne, UK) at a dilution of 1:100. After the overnight incubation with the primary antibody, the sections were first reacted for 30 min with biotinylated antimouse IgG (Vector Laboratories, CA) diluted 1:200 in PBS containing 1% normal horse serum, and then for 30 min in avidin-biotinylated peroxidase complex (Vectastain ABC Kits, Vector). Between each step in the staining procedure, sections were rinsed with PBS for 3 × 5 min. Staining was visualized with 3-amino-9-ethyl-carbazole (A5754; Sigma, St. Louis, MO), 0.2 g/ml in 0.5 M acetate buffer containing 0.01% hydrogen peroxide (pH 5.0) for 15 min. Subsequently the sections were washed thoroughly in tap water for 10 min, counterstained with Meyer's haematoxylin for 15–60 sec, washed, and mounted in aqueous mounting media (Aquamount; BDH, Poole, UK). All the steps of this procedure were carried out at room temperature. The principle of the HuR immunostaining protocol was similar to that of cyclin A and has been described previously.15 A monoclonal HuR antibody (1:10 000 dilution, 19F12) was the kind gift of Clonegene, LLC (Hartford, CT). Sections treated with PBS (instead of specific antibody) served as negative controls.
Cyclin A immunoreactivity was measured by estimating the number of tumor cells and the percentage of positive nuclei using a medium power (×20) objective in areas where the staining was most intense. At least 5 fields were scored. A value less than 5% was considered negative, and one equal to or greater than 5% was considered positive.17, 18 Cyclin A immunostaining was scored in a consensus manner without preliminary knowledge of clinical data (J.M. and S.N.). HuR immunoreactivity was scored by 2 independent interpreters (J.M. and A.R.) without information as to clinicopathological factors. The final consensus score, case-by-case, was discussed and determined in a common session. We observed immunoreactivity in the nuclei and the cytoplasm as follows: nucleus only, cytoplasm only, nucleus and cytoplasm and totally negative. From these subgroups we categorized those which werenucleus-negative and -positive, and cytoplasm-negative and -positive.
The associations between Cyclin A and HuR scores and clinicopathological variables were assessed by χ2- or Fisher's exact test. Univariate survival analysis was calculated by the Kaplan-Meier method, and difference between groups was assessed by the logrank test or logrank test for trend. Deaths other than those due to gastric cancer were censored. Multivariate survival analysis was performed by the Cox proportional hazards model. A p-value <0.05 was considered significant, and as an inclusion criteria for multivariate analysis.
Cyclin A immunoreactivity was negative in 146 of the 325 (45%) specimens, and positive in 179 (55%). Positive staining was observed in the nuclei of the tumor tissue and seldom was there any cytoplasmic immunoreactivity (Fig. 1). The immunoreactivity of HuR was positive in the cytoplasm in 127 of the 316 (40%) specimens and in the nuclei in 279 (88%).
Associations of cyclin A with clinicopathological factors and HuR
High expression of cyclin A was associated with old age, high stage, proximal location of the tumor, intestinal subtype, and noncurative resection (Table I). Advanced penetration depth of the tumor and nodal metastases were also associated with high cyclin A level, whereas distant metastases were not. According to our previous data, 198 of 309 (64%) of the specimens had positive cytoplasmic HuR immunoreactivity, and 273 (88%) had nuclear positivity.16 Cytoplasmic HuR immunoreactivity associated with cyclin A, (p < 0.001, χ2 = 12.0) (Table II), but nuclear did not (p≤ 1.0).
|Immunoreactivity||HuR cytoplasm neg.||HuR cytoplasm pos.||Total|
|Cyclin A (low)||94||39||133|
|Cyclin A (high)||90||86||176|
High cyclin A immunoreactivity was associated with poor survival (p < 0.0001, Fig. 2). In diffuse type cancers, cyclin A positivity was associated with poor survival (n = 184, p < 0.0001, χ2 = 16.0), whereas patients with intestinal type cancer expressing high cyclin A had a trend for unfavorable prognosis (n = 141, p = 0.065, χ2 = 3.4). Of the other parameters, old age, high stage, advanced penetration depth, nodal or distal metastases, proximal location of tumor, large tumor size and non-curative surgery were markers of poor survival (Table III). In multivariate analysis, cyclin A and stage remained independent prognostic factors (Table IV).
|Clinicopathological variable||Patients||Cumulative 5-year survival %||χ2||p value|
|Cyclin A immunoreactivity||16.80||<0.0001|
|Penetration depth (pT)||133.61||<0.0001|
|Lymph node metastases||165.14||<0.0001|
|Data not available||1|
|Data not available||27|
|Tumor size (cm) median||60.75||<0.0001|
|Data not available||6|
|Intent to cure||165||58|
|Data not available||18|
|Covariate||p value||RH||CI (95%)|
|Cyclin A immunoreactivity||0.0163||1.43||1.13–1.73|
|Tumor size (<5 cm)||NS|
We are, to our knowledge, the first to report cyclin A immunoreactivity to be an independent prognostic factor in gastric cancer. In other cancers, the prognostic role of cyclin A is controversial. High cyclin A immunoreactivity in esophageal squamous cell carcinoma is associated with tumor progression and unfavorable outcome.19 Handa et al. showed cyclin A overexpression to be associated with poor outcome and to be an independent prognostic factor in colorectal cancer.20 Bahnassy et al.21 and Nozoe et al.22 have reported similar results concerning colon cancer. In contrast, Li et al. showed that cyclin A expression gradually increases in the colon adenoma-carcinoma sequence, but patients with high cyclin A immunoreactivity had better survival than those with reduced cyclin A expression.23 They also described invasive primary tumors and hepatic metastases to have low immunoreactivity rates of cyclin A. The findings of Bondi et al. are congruent with the report of low expression of cyclin A, showing a significant association with unfavorable prognosis in colon cancer.18 It has been suggested that a decrease, rather than increase, in cyclin A in cells promotes tumorigenesis, and once the tumor has developed, high levels of cyclin A indicate the high proliferation rate,2 which could explain this inconsistency of those studies. Only a few reports concerning cyclin A expression in gastric cancer have been made, but no correlation with survival was evident.8 However, the majority of previous works particularly emphasize that overexpression of cyclin A is a marker for poor prognosis in cancer.24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34
There are 2 known isoforms of cyclin A. Cyclin A1 is the embryonal form of the protein, and its role is limited to male meiosis.35 Cyclin A1 is also expressed in normal human testis and brain tissues, and in myeloid leuchemia cell lines.36 In contrast, cyclin A2 is present in proliferating somatic cells, and the cyclin A2 gene is embryonic lethal if disrupted.37 Thus, synthesis of cyclin A2 and its binding to CDKs during cell cycle are widely characterized and can be considered essential in carcinogenesis and cancer prognostics, whereas the role of cyclin A1 remains less evident in solid tumors.38 It is unknown whether the isoforms have any confounding effect in gastric cancer prognostics, but we consider it unlikely.
We found, in addition to reduced survival, that a high level of cyclin A expression was associated with several clinicopathological factors reflecting aggressive properties of the disease, such as high stage, tumor penetration and nodal metastasis. Old age, proximal location of the tumor, intestinal type, and noncurative surgery were also associated with high cyclin A immunoreactivity. Since regulation of cyclin A expression is partly mediated via mRNA stabilization, we studied whether there is a link between cyclin A and the mRNA stabilizing factor HuR. To our knowledge, we are the first to report the association of cyclin A expression and HuR expression in any clinical cancer material. When cytoplasmic HuR was positive, the frequency of cyclin A negative specimens was relatively low. This is supported by data from colon cancer cell culture experiments, where reduction in HuR leads to decreased expression of cyclin A.13 We thus suggest that although cytoplasmic HuR may not be the only determinant of cyclin A expression, cytoplasmic presence of HuR may facilitate cyclin A expression on gastric cancer.
Surgery is the only effective treatment to cure gastric cancer, but certain patients benefit from adjuvant therapy. Kawashima etal. reported that cytotoxicity of 5-FU in human cancer cell lines correlates with cyclin A.39 In nonsmall-cell lung carcinomas, expression of cyclin A correlates with response of doxorubicin invitro,40 and patients with soft tissue sarcoma41 or head and neck squamous cell carcinoma42 expressing high levels of cyclin Ashow a better response to chemotherapy. These data suggest that cyclin A overexpression may prove useful in selecting those gastric cancer patients who may benefit from chemotherapy. However, to support this theory, more experimental studies and clinical trials should be performed.
In conclusion, we report that cyclin A is associated with reduced survival and clinicopathological variables in gastric cancer. Our results also suggest that cyclin A is associated with cytoplasmic expression of HuR, which may be one of the factors controlling cyclin A expression in gastric cancer.
Technical assistance of Ms. Elina Laitinen, Mr. Päivi Peltokangas, Ms. Elina Aspiala and Ms. Tuija Hallikainen is warmly acknowledged, as well as language revision by Ms. Carol Norris.