Expression of cyclins A and D and p21(waf1/cip1) proteins in renal cell cancer and their relation to clinicopathological variables and patient survival

We have studied 118 renal cell carcinomas to analyse the expressions of cyclins A and D1 and p21(waf1/cip1), and their relationship to clinical and histopathological parameters as well as to clinical outcome. Cyclins A and D1 and cyclin-dependent kinase inhibitor p21(waf1/cip1) were not expressed in normal renal tissue. Staining signals of cyclin D1 and p21(waf1/cip1) were always nuclear but cyclin A was also expressed in the cytoplasm of the tumour cells. The mean (range) fractions of cyclin A, cyclin D1 and p21(waf1/cip1)-positive tumour cells were 2.2% (range 0–20%), 23.3% (range 0–90%) and 6.8% (range 0–70%) respectively. The expression of cyclin A was related to venous invasion, high nuclear grade, high mitotic rate, high Ki-67 and high PCNA expressions (P ≤ 0.006 for all). The expression of cyclin D1 was linked with age over 65 years, low nuclear grade and high p53 expression (P ≤ 0.05 for all). An inverse correlation was present between p21(waf1/cip1) and cyclin D1 (P = 0.011). Cyclin A predicted survival in the entire study group (P = 0.0014), in T1–4/N0–2/M0 (P = 0.0007) and in T1–2/N0/M0 tumours (P = 0.0007). Cyclin A was also a powerful predictor of disease-free survival in T1–4/N0/M0 (P = 0.0027) tumours (P = 0.0007). Cyclin D1 and p21(waf1/cip1) were not significantly related to survival or disease-free survival in any of the groups. In the entire material the independent prognostic factors were the presence of distant metastases (relative risk (RR) 5.16, P < 0.001), T category (RR 2.68, P < 0.001), Ki-67 expression (RR 1.02, P = 0.026) and cyclin A expression (RR 1.12, P = 0.001). The independent predictors in T1–4/N0/M0 tumours were T-category (RR 2.67, P = 0.001) and cyclin A (RR 1.21, P < 0.001), and in T1–2/N0/M0 tumours the only significant predictor was cyclin A (RR 1.19, P = 0.0002). In renal cell carcinoma, cyclin A is a powerful and independent prognostic factor in all clinical stages of the disease, whereas cyclin D1 and p21(waf1/cip1) have no prognostic value. © 1999 Cancer Research Campaign

The main therapy in renal cell carcinoma (RCC) is operative treatment but the operative results are variable (Sweeney et al, 1996;Giberti et al, 1997). In a proportion of the cases, adjuvant treatment may be helpful to improve the survival of patients. To identify the group of patients who may benefit from additional treatment modalities, new accurate prognostic factors are urgently needed. Today the TNM classification and nuclear grade are the main factors on which the treatment decisions are based, but better prognostic factors are under research to tailor the therapy individually. The knowledge and understanding of the biological nature of RCC has increased and new promising prognostic indicators have been already identified. These markers include indicators of cell proliferation (Aaltomaa et al, 1997;Papadopoulos et al, 1997), cell adhesion (Koga et al, 1997;Paul et al, 1998), markers involved directly in the regulation of cell growth (Hofmockel et al, 1997;Shiina et al, 1997) and growth-stimulating or suppressing gene products (Hofmockel et al, 1997;Shiina et al, 1997).
The suppressor gene p53 is frequently mutated in different cancers and it is able to stop the cell cycle at G1 phase and also to induce apoptosis. The recent reports suggest that p21 (waf1/cip1) is a main mediator of p53 tumour suppressor gene effects (EI Deiry et al, 1995;Naka et al, 1998). Elevated levels of p21 (waf1/cip1) protein in response to both p53-dependent and -independent signals mediate the cell cycle arrest predominantly at the G1 phase of the cell cycle (Zeng et al 1996). p21 (waf1/cip1) exerts this effect by inhibiting the cyclin-dependent kinases (cdk) that are required to drive the cell division cycle. While active cdk are found in normal cells in a quaternary complex with cyclins, proliferating cell nuclear antigen and p21 (waf1/cip1) , more than one molecule of p21 (waf1/cip1) per complex results in kinase inhibition (Waga et al, 1994;Zhang et al, 1994). In experimental analyses of human bladder cancer cell lines p21 (waf1/cip1) is able to induce apoptosis, but the exact role of p21 (waf1/cip1) in apoptosis is not completely understood (Kawasaki et al, 1998). In RCC the significance of p21 (waf1/cip1) protein in tumour differentiation and patient survival has not been previously studied. Cyclin A and D1 are protein kinases related to control of the cell cycle and cell proliferation (Donnellan and Chetty, 1998). Cyclins act together with the cdk by phosphorylating the main substrates responsible for the cell cycle regulation (Donnellan and Chetty, 1998). Phosphorylation of the retinoblastoma (Rb) tumour suppressor protein (pRB) by the complexes of cyclins and cdk makes Rb protein active to induce genes controlling cell proliferation, thus being a part of the control of the cell cycle machinery (Strauss et al, 1995;Donnellan and Chetty, 1998). Cyclin D1 is most active in G1 resting phase before S-phase by regulating the DNA synthesis, by phosphorylating the pRB and simultaneously activating its function (Strauss et al, 1995;Donnellan and Chetty, 1998). Cyclin A regulates the cell Expression of cyclins A and D and p21 (waf1/cip1) proteins in renal cell cancer and their relation to clinicopathological variables and patient survival cycle also through the pRB activation (Donnellan and Chetty, 1998). The overexpression of cyclins has been related to high proliferation rate of the tumour cells and to other unfavourable prognostic factors, although the results are not uniform (Aaltomaa et al, 1999a;Donnellan and Chetty, 1998;Ishikawa et al, 1998;Maeda et al, 1998;Volm et al, 1998). The gene amplification and protein expressions of cyclins A and D1 have been previously studied in several human epithelial neoplasms to establish their possible prognostic significance in clinical human cancer (Barbareschi et al, 1997;Michalides et al, 1997;Aaltomaa et al, 1999a;Donnellan and Chetty, 1998;Maeda et al, 1998;Volm et al, 1998). At present there are no reports available on the prognostic value of the expressions of cyclins A and D1 in RCC.
The aims of our study were to analyse the expressions of cyclins A and D1 as well as p21 (waf1/cip1) protein in RCC and to compare the expressions of these proteins to other clinical and histopathological parameters, and to establish the prognostic value of these proteins in RCC patients.

PATIENTS AND METHODS
The study includes 118 RCC patients treated between 1968 and 1985 and followed up to 1991 at the Department of Surgery, Kuopio University Hospital, Finland. The cohort was not entirely consecutive since adequate biopsy specimens for immunohistochemistry were not available in all cases. The estimated total number of diagnosed RCCs during the enrolment period was about 200 cases, but an operation was carried out in only about half of the cases. The mean (range) age of the patients at diagnosis was 61 (range 22-82) years and the mean (range) follow-up time was 9.7 (range 5.3-20.2) years. The follow-up of patients after therapy lasted until death or to the year 1991. The diagnosis, clinical staging, treatment and clinical follow-up were carried out mainly by two urologists. The diagnosis was based on the results of routine laboratory tests and intravenous (i.v.) urography, and on ultrasonography and computerized tomography (CT) during the last years of follow-up. The histological confirmation was needed before the diagnosis was settled. In addition, the bone chart, chest X-ray and ultrasonography were done to detect the possible metastasis. During the first 2 years the follow-up was done on every 3rd month and thereafter every 6th month up to 5 years, and after that controls were started once a year. During the follow-up, routine laboratory tests and chest X-rays were done on regular basis, and ultrasonography, i.v. urography, CT and the bone chart when they were indicated. There were 114 cases treated with radical nephrectomy, in four cases a partial nephrectomy was done. The causes of death were verified from the patient files, death certificates and from the files of the Finnish Cancer Registry.

Histological methods
Five micrometer-thick serial sections were cut from the paraffinembedded surgical specimens and stained with haematoxylin and eosin. The samples were examined in a blinded manner. Tumours were categorized into four nuclear grades as described in detail in the previous literature (Syrjänen and Hjelt, 1978). The mitotic figures were identified, usually at the tumour periphery at the areas of the most proliferative and invasive growth. The counting of mitotic figures was done by using an objective magnification of 40 × (field diameter 490 µm) and the volume corrected mitotic index (M/V) method was used (Haapasalo et al, 1989). The M/V index expresses the number of mitotic figures mm -2 of neoplastic epithelium in the section. The mean M/V index was 6.9 (standard deviation (s.d.) range 0-44) mm -2 .

Cyclin A and D1 and p21( waf1/cip1 ) immunohistochemistry
Cyclin A (Aaltomaa et al, 1998a) was demonstrated by a routine immunohistochemical method, which is similar to that described in connection with Ki-67 immunostaining except that microwave pretreatment was not used in cyclin A immunohistochemistry. The antibody was purchased from Novocastra Laboratories (Newcastle upon Tyne, UK) and it was used at a dilution of 1:200. Cyclin D1 (Aaltomaa et al, 1999a) and p21 (waf1/cip1) (Lipponen et al, 1998) were demonstrated by using the same staining procedure as described in connection with Ki-67 immunohistochemistry (Aaltomaa et al, 1997). Both of the antibodies were purchased from Novocastra Laboratories (Newcastle upon Tyne, UK) and they were used at a dilution of 1:100 (cyclin D1) and 1:20 (p21 (waf1/cip1) ). The expressions of cyclins and p21 (waf1/cip1) (fraction of positive tumour cells) were evaluated in the entire section (magnification 400 ×). The expression of cyclin D1 was considered positive only when distinct strong nuclear positivity was present. Faint expression of cyclin D1 was present in most of the cancers, but it was not included in the scoring process as recommended in the previous literature (Gillet et al, 1996). The mean (s.d.) fractions of cyclin A-and D1-positive nuclei were 2.2% (s.d. 4.1%) and 23.2% (s.d. 28.7%) respectively. The mean (s.d.) fraction of positive nuclei for p21 (waf1/cip1) was 6.8% (s.d. 10.5%).

Ki-67 immunohistochemistry
For immunohistochemical demonstration of Ki-67 protein, 5-µm sections were deparaffinized, rehydrated and washed for 5 min with phosphate-buffered saline (PBS). Thereafter the sections were rinsed in distilled water and heated in a microwave oven for 2 × 5 min in 0.01 M citrate buffer (pH 6.0). After that the slides were rinsed in Tris-buffered saline (pH 7.4). Endogenous peroxide was blocked by 3% hydrogen peroxide for 5 min followed by a wash for 5 min with PBS. The tissue sections were incubated with the monoclonal anti-Ki-67 protein (MIB1, Dianova Marseille, France) antibody diluted at 1:100 in PBS. Sections were washed twice for 5 min with PBS, incubated for 20 min with biotinylated secondary antibody (Vectastain ABC Elite Kit, Vector Laboratories, CA, USA) diluted at 1:200 in PBS. Slides were washed twice in PBS for 10 min and incubated for 20 min in preformed avidin-biotinylated peroxidase complex (Vectastain ABC Elite Kit, Vector Laboratories, CA, USA). Sections were washed twice for 5 min with PBS, developed with diaminobenzidine tetrahydrochloride substrate (Sigma Chemical Co., St Louis, MO, USA), slightly counterstained with Mayer's haematoxylin, dehydrated, cleared and mounted. Normal human tonsil was used as a positive control. The fraction of positively stained nuclei was estimated in the area of the tumour that contained the highest fraction of Ki-67-positive cells. The mean (s.d.) fraction of Ki-67 positive nuclei was 14.2% (s.d. 15.7%).

p53 and PCNA immunohistochemistry
The expression of these proteins was detected by a routine immunohistochemical method as detailed in previous literature (Lipponen et al, 1994). The antibody for p53 (CM1, Novocastra Laboratories, Newcastle upon Tyne, UK) was used at a dilution of 1:1200. The antibody for PCNA (PC10) was purchased from DAKO (Golstrup, Denmark) and it was used at a dilution of 1:100. The fraction of positive cells was analysed from the areas with the highest fraction of positive cells as detailed before (Lipponen et al, 1994). The mean (s.d.) fraction of p53-positive cells was 1.2% (s.d. 7.6%) and of proliferating cell nuclear antigen (PCNA) 29.3% (s.d. 30.4%).

Statistical methods
In statistical calculations, the SPSS-X was used in an IBM computer and the statistical tests used are indicated in the results when appropriate. Univariate survival analysis was done by the SPSS-X using the Kaplan-Meier method with statistics by Gehan. Multivariate survival analysis (Cox's analysis) was done in a stepwise manner. Survival analysis included (as an event) only the deaths due to renal cell carcinoma. During the follow-up, 68 patients died of renal cell carcinoma and three died from other causes. When the disease-free survival was analysed, only M0 cases (n = 82) were included in the analysis. The group limits (when categorized data is used) for Ki-67, p53 and PCNA were set so that these parameters showed the highest possible correlation to cyclin A, D and p21 (waf1/cip1) expression in Table II.
The expression of cyclins A and D1 were not related to sex, T-category, M-category or tumour size (P > 0.05), but the expression of cyclin A was related to venous invasion (P = 0.04) and cyclin D to age over 65 years (P = 0.03). The relationship between the histopathological parameters and cyclin A and D1 is shown in Table 1. The expression of p21 (waf1/cip1) protein was not related to any of the parameters shown in Table 1. The correlation coefficients and their significance between the different histopathological tumour markers are shown in Table 2. Cyclin A predicted survival in the entire study group (P = 0.0014, Figure 2), in M0 (P = 0.0007, Figure 3) and in T1-2/N0/M0 cases (P = 0.0007, Figure 4). It was also a powerful predictor of disease-free survival in T1-4/N0/M0 (P = 0.0027) and in T1-2/N0/M0 tumours (P = 0.0007). Cyclin D1 and p21 (waf1/cip1) were not significantly related to patient survival or disease-free survival in any of the analysed subgroups. The independent prognostic factors of survival are shown in Table 3.

DISCUSSION
Cyclins are closely related to cell cycle by acting together with cdk in phosphorylating the Rb protein. Type D1 cyclin is mostly expressed during the first gap phase (G1) of the cell cycle and it is a major factor in the timing process of starting DNA synthesis in mammalian cells (Strauss et al, 1995;Kato, 1997;Donnellan and Chetty, 1998). On the contrary, cyclin A is expressed mainly in late S and G2 phases and is degraded during mitotic phase before  M/V index, mitotic index: PCNA, proliferating cell nuclear antigen. Table 3 The independent prognostic factors (final step analysis) of survival in multivariate analysis, which included all the analysed prognostic parameters. Some of the parameters were not available in all cases and accordingly the total number of cases included is lower than in the original cohort Multivariate analyses included following variables: T-classification, node status, presence of metastasis, nuclear grade, mitotic rate, Ki-67, cyclin A and D1, p21 (waf1/cip1) , p53, proliferating cell nuclear antigen (PCNA), sex and age. RR, risk ratio; Cl, 95% confidence interval. the metaphase (Donnellan and Chetty, 1998). The increased expression of cyclins is related to disregulated or accelerated cell cycle as a result of gene amplification, chromosomal translocation, down-or up-regulated expression (Naitoh et al, 1995;Barbareschi et al, 1997;Aaltomaa et al, 1999a;Volm et al, 1998). So far, no previous study has compared the relationships between the expressions of cyclins and p21 (waf1/cip1) , p53 as well as Ki-67 expressions in RCC. Consequently, the prognostic role of cyclins in RCC has not yet been defined. Although in other carcinomas the relationship between TNM classification and cyclins is obvious (Michalides et al, 1997;Aaltomaa et al, 1999a), we could not find any correlation between the expression of cyclin A and TNM categories. Differentiation grade was significantly associated with the fraction of cyclin Apositive cells, which is in line with the results from prostate cancer as well (Aaltomaa et al, 1999a). The expression of cyclin A was also related to many indicators of cell proliferation such as Ki-67 and PCNA expressions and mitotic rate, a finding which is in accordance with previous reports (Marshall et al, 1996;Aaltomaa et al, 1999a). The staining signal of cyclin A was independent of p53 and p21 (waf1/cip1) expressions, which suggests that these latter proteins may be of minor importance in cell cycle control in RCC. Additional support for this suggestion comes from the study (Papandreou et al, 1997) in which the p21 (waf1/cip1) gene was rarely mutated in RCC.
Cyclin D1 was not related to TNM classification in RCC in the present study. A high fraction of cyclin D1-positive cells was linked with low cell proliferation and well-differentiated tumours.
There are no reports available in the literature to verify our results in RCC, but in breast cancer the expression of cyclin D1 is related to sex steroid receptor positivity and favourable histological signs (Gillet et al, 1996;Barbareschi et al, 1997). Also, in bladder cancer, the initial results suggest that cyclin D1 might be associated with low histological grade . The above results in bladder and breast cancer are in accordance with our results in RCC. However, the results are variable, since the close relationship between overexpression of cyclin D1 and malignant cellular prognostic features has been observed in other cancers (Nakamura et al, 1997;Pignataro et al, 1998). In gastric and colon cancer the strong expression of cyclin D1 is related to lymph node involvement (Nakamura et al, 1997;Maeda et al, 1998), and in ovarian cancer to tumour malignancy (Barbieri et al, 1997). In addition, our analysis revealed a positive correlation (Table 2) between the expression of p21 (waf1/cip1) and cyclin D1. This association is most probably due to reduced cell proliferation of cyclin D1-positive tumours and possibly to growth inhibitory functions of p21 (waf1/cip1) .
In RCC, the staining signal of p21 (waf1/cip1) seems to be independent of p53 expression. Indeed, p53 independent expression of p21 (waf1/cip1) has been reported in several tumour types, including colon, endometrium, prostate and bladder cancer (Backe et al, 1997;Yasui et al, 1997;Aaltomaa et al, 1999b;Lipponen et al, 1998). Unexpectedly, p21 (waf1/cip1) was not associated with the expression of PCNA or other indicators of accelerated cell proliferation, although in cells p21 (waf1/cip1) exists as a protein complex together with PCNA, cyclin and cdk. The data relating p21 (waf1/cip1) to cell proliferation are rather variable (Backe et al, 1997;Aaltomaa et al, 1999b;Lipponen et al, 1998). In bladder cancer, p21 (waf1/cip1) was related to high cell proliferation, but in another study no significant relationship could be found between these variables (Lipponen et al, 1998). These conflicting results are most probably due to altered stoichiometry of these proteins in malignant cells, and the metabolism of p21 (waf1/cip1) may be disturbed because of an altered protein structure. Also, it is possible that p21 (waf1/cip1) -positive tumour cells represent a subpopulation of cells that have withdrawn from the cell cycle amid the otherwise rapidly proliferating tumour cells (El Deiry et al, 1995).
Our survival analysis showed that p21 (waf1/cip1) had no prognostic value in RCC. Similar results have been reported in bladder, prostate, endometrium and hepatocellular cancer (Backe et al, 1997;Byrne et al, 1997;Aaltomaa et al, 1999b;Lipponen et al, 1998;Naka et al, 1998). On the contrary, in oesophageal and colon cancer, as well as in squamous cell carcinomas of the head and neck, the strong expression of cyclin D1 was related to unfavourable prognosis (Ishikawa et al, 1998;Maeda et al, 1998). In addition, in colorectal adenocarcinoma and in superficial bladder cancer the overexpression of cyclin D1 predicts early recurrence (Shin et al, 1997;Maeda et al, 1998). Moreover, cyclin D1 overexpression is related to favourable prognosis in breast cancer (Gillet et al, 1996). In RCC we found no correlation between the expression of cyclin D1 and survival or recurrencefree survival, whereas cyclin A was a powerful predictor of patient outcome, and the independent prognostic value of cyclin A was confirmed also in multivariate analyses of the various subgroups of the tumours. If the cases were categorized into three groups -0%, 1-5% and over 5% -of cyclin A-positive cells, the corresponding survival rates at 10 years were 60%, 30% and 10% respectively (P < 0.0001). This clearly shows that cyclin A expression can be used in categorizing RCCs into prognostic groups.