SEARCH

SEARCH BY CITATION

Keywords:

  • peroxisome proliferator-activated receptor gamma;
  • PPARγ;
  • renal cell carcinoma;
  • underexpression

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

Aim:  To examine peroxisome proliferator-activated receptor gamma (PPARγ) mRNA expression quantitatively in human renal cell carcinoma (RCC) cell lines and RCC tissue, as well as in corresponding normal kidney tissue.

Methods:  We examined PPARγ mRNA expression quantitatively in six human RCC cell lines by real-time reverse transcription–polymerase chain reaction. In addition, we evaluated the relationship between cell growth inhibition by PPARγ ligands and the level of PPARγ mRNA expression. We compared the expression of PPARγ mRNA in 47 RCC tissues with that in corresponding normal kidney tissue, and investigated the relationship between clinicopathological features and the level of PPARγ mRNA expression.

Results:  Among the six RCC cell lines, five showed decreased PPARγ mRNA expression. There was no relationship between the inhibitory effects of PPARγ ligands and PPARγ mRNA expression levels. Of the tissues from 47 RCC patients, 25 (53%) showed decreased expression of PPARγ mRNA compared to corresponding normal kidney tissue, and one was equivalent to normal tissue. These patients had distant metastasis at diagnosis more frequently than the remaining patients with high expression. There was also a trend for these patients to have a higher stage.

Conclusions:  Most RCC cell lines showed decreased expression of PPARγ mRNA. However, the level of PPARγ mRNA expression did not affect cell growth inhibition by PPARγ ligands. More than half of the tissues from RCC patients had low expression of PPARγ mRNA, and such carcinomas might have more aggressive behavior.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

Peroxisome proliferator-activated receptor gamma (PPARγ) belongs to the steroid receptor superfamily of ligand-activated transcription factors.1,2 In addition to its important role in adipose differentiation, PPARγ is involved in a variety of normal physiological processes such as lipid metabolism, glucose metabolism, inflammatory responses, and macrophage differentiation.3–5 Moreover, PPARγ has been shown to be expressed in various malignant cells, and exert antiproliferative effects by its ligands, which include natural ligands (15-deoxy-Δ12,14-prostaglandin J2 (15-PGJ2), and unsaturated fatty acid) and synthetic ligands belonging to the thiazolidinedione class of antidiabetic drugs (troglitazone, pioglitazone, ciglitazone, and so on).6,7 Recently, we have reported that PPARγ activation by 15-PGJ2 and pioglitazone induces growth inhibition, apoptosis, and antiangiogenesis in human renal cell carcinoma (RCC) cells.8 These findings suggest that PPARγ may be a novel target in the treatment of RCC, which is one of most difficult cancers to treat because of its resistance to conventional therapies such as chemotherapy and radiotherapy.

On the basis of the above-mentioned theory, it could be expected that higher expression of PPARγ in a tumor would result in more effective PPARγ activation by its ligands for inhibiting cell growth. However, recent studies have shown that PPARγ expression is down-regulated in some cancers, including esophageal, lung, and thyroid cancers,9–11 suggesting that PPARγ could be a candidate tumor suppressor gene. Indeed, functional mutations of PPARγ were identified in 4 of 55 colorectal cancers.12 Furthermore, decreased expression of PPARγ is correlated with poor prognosis in patients with esophageal and lung cancer.9,10 Thus, it may be important to know the status of PPARγ expression before clinical treatment.

In this study we examined the expression of PPARγ mRNA quantitatively in RCC cell lines and RCC tissue. Furthermore, we investigated correlation of the level of PPARγ mRNA expression with clinicopathological features in RCC patients.

Materials and methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

Cell lines, cultures and reagents

Human renal cell carcinoma cell lines SMKT-R-1, 2, 3 and 4, which were established in our laboratory, were maintained in minimal essential medium with d-valine modification medium containing 10% fetal bovine serum. Each cell line was derived from a primary lesion of an individual patient with RCC. All cell lines were histopathologically proven to be of conventional RCC origin.13,14 Human renal cell carcinoma cell lines ACHN and Caki-1 were purchased from the American Type Culture Collection (Rockville, MD, USA), and were cultured in RPMI 1640 medium with 10% fetal bovine serum. Pioglitazone was kindly provided by Takeda Chemical Industries (Osaka, Japan). 15d-PGJ2 was obtained from Calbiochem (La Jolla, CA, USA). Pioglitazone and 15d-PGJ2 were dissolved in 0.1% in dimethylsulfoxide.

Tissue samples

Samples were obtained from 47 RCC patients who had undergone operations at Sapporo Medical University Hospital from 2001 to 2004. Clinicopathological features of 47 RCC patients are summarized in Table 1. All patients provided written informed consent. The tumor grade was classified into grades 1–3, depending on the degree of nuclear anaplasia according to the UICC-TNM classification.15 The pathological stage was determined as well, according to the classification. Viable portions of tumors were selected carefully for analysis. The cortex portion of corresponding normal kidney tissue was used as a comparative control. All samples for DNA sequencing and reverse transcription–polymerase chain reaction (RT–PCR) were frozen immediately in liquid nitrogen and stored at −80°C until analysis.

Table 1.  Patients’ characteristics
CharacteristicsNumber of patients (%)
Age (years)36–85 (Mean, 63.3)
Gender
 Male34 (72)
 Female13 (28)
Pathological stage
 pT1a13 (28)
 pT1b18 (38)
 pT2 4 (8)
 pT312 (26)
Grade
 1 6 (13)
 230 (64)
 311 (23)
Histology
 Clear cell43 (92)
 Chromophobe 2 (4)
 Spindle 1 (2)
 Bellini duct 1 (2)
Metastasis at diagnosis
 No38 (81)
 Yes 9 (19)
Follow up (months)3.2–35.3 (Median, 21.4)

Real time reverse transcription–polymerase chain reaction

Total RNA was extracted from the RCC cell lines, the RCC tissues and corresponding normal kidney tissues, using Trizol Reagent (Invitrogen, Carlsbad, CA, USA). Three micrograms of total RNA was reverse transcribed into cDNA. For real-time PCR, each cDNA (3 µL) was amplified in triplicate with the use of a QuantiTect SYBR Green PCR Kit (Qiagen, Hilden, Germany) and then detected using an ABI PRISM 7700 Sequence Detection System (Applied Biosystems, Foster City, CA, USA). PCR conditions were 50°C for 2 min and 95°C for 15 min, followed by 45 cycles of 95°C for 15 s, 57°C for 40 s and 72°C for 40 s. β-Actin was used as the endogenous RNA control. The primers for real-time PCR were, PPARγ sense primer, 5′-AGTGGG GATGTCTCATAATGCC-3′, and antisense primer, 5′-AGCTCAGCGGACTCTGGATTC-3′;16β-actin sense primer, 5′-GGAAAAGATGAGTATGCCTG-3′, and antisense primer, 5′-TTCACTCAATCCAAATGCGG-3′, which yielded PCR products of 117 bp and 135 bp, respectively. Specificity of the PCR was checked by analysing melting curves. Relative mRNA levels were calculated using the ΔΔCt method by comparing the PCR cycle threshold of cDNA of PPARγ and that of β-actin.17

DNA sequencing

Genomic DNA was isolated from six RCC cell lines. PCR was performed for amplification of PPARγ exon 3 and exon 5 using a GeneAmp RNA PCR kit (Applied Biosystems). The primers used were as follows: PPARγ exon 3 sense primer, 5′-GCTTCCATGTGTCATAAAGACTTAA-3′, and antisense primer, 5′-GGGCTGCAGCTATAATGAG-3′; PPARγ exon 5 sense primer, 5′-CGACCAAGTAA CTCTCCTCA-3′, and antisense primer, 5′-CCATCATC CCACCCTCTTTC-3′.12 Amplified products were purified with a QIAquick PCR Purification Kit (Qiagen), cloned using a TOPO TA Cloning Kit (Invitrogen), sequenced using BigDye Terminator v3.1 Cycle Sequencing Kit (Applied Biosystems) and ABI 3100 Genetic Analyzer, according to the manufacturers’ instructions. In each case, 10 clones were sequenced.

Cell proliferation assay

To evaluate the relationship between PPARγ mRNA expression levels and the inhibitory effects of PPARγ ligands in human RCC cell lines, a modified 3-(4,5-dimethylthiazol-2-thiazolyl)-2, 5 -diphenyltetrazolium bromide assay was performed as described previously.8 In brief, the cells were seeded on a 96-well tissue culture plate (Becton Dickinson, Franklin Lakes, NJ, USA), and incubated with pioglitazone or 15d-PGJ2 at various concentrations (5, 10, 25, and 50 µmol/L) for 48 h. Cell viability was measured using a Cell Counting Kit-8 (Dojindo, Kumamoto, Japan).

Statistical analysis

Statistical analysis was carried out with SPSS 11.0 for Windows (SPSS, Chicago, IL, USA). The Mann–Whitney U-test, Fisher's exact test or χ2 test were used to evaluate the correlation between the expression of PPARγ mRNA and clinicopathological features. The result was considered significant when the P-value was <0.05.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

Quantitative analysis of PPARγ mRNA expression in human renal cell carcinoma cell lines

First, we examined PPARγ mRNA expression in six human RCC cell lines quantitatively by real-time RT–PCR. PPARγ mRNA was detectable in all the cell lines at various levels. All the RCC cells lines except ACHN showed decreased PPARγ mRNA expression compared to three non-corresponding normal kidney tissues, which were used to calibrate the relative expression of PPARγ expression. Relative expression levels of PPARγ mRNA in SMKT-R-1, R-2, R-3, R-4, and Caki-1 were 0.19, 0.74, 0.66, 0.62, and 0.23, respectively. Only ACHN cells had higher expression (2.46) of PPARγ mRNA. To investigate whether decreased PPARγ mRNA expression was due to PPARγ gene mutation, we screened the six RCC cell lines for mutation of exon 3 (DNA-binding domain) and exon 5 (ligand-binding domain) of the PPARγ gene by DNA sequencing analysis. No mutations were detectable in any of the six RCC cell lines.

Relationship between PPARγ mRNA expression levels and cell growth inhibition by PPARγ ligands in human renal cell carcinoma cell lines

Similar to a previous report showing that both pioglitazone and 15d-PGJ2 inhibited cell growth in SMKT-R-1, R-2, R-3, and R-4 cells,8 these ligands also suppressed cell proliferation in Caki-1 and ACHN cells in a dose-dependent manner (Fig. 1a,b). There was no significant difference in regard to cell growth inhibition at each concentration of the ligands between Caki-1 cells with low PPARγ mRNA expression and ACHN cells with high PPARγ mRNA expression.

image

Figure 1. (a,b) Effects of the PPARγ ligands on the growth of ACHN (◊) and Caki-1 (□) cells. These cells were treated with the indicated concentrations of pioglitazone and 15d-PGJ2 for 48 h, and then were measured by the MTT assay. Values represent the percentage of inhibition of control culture from three independent experiments (n = 9). Data represent mean ± SD. *P < 0.001 versus control.

Download figure to PowerPoint

Quantitative analysis of PPARγ mRNA expression in human renal cell carcinoma tissue

To examine whether down-regulation of PPARγ mRNA expression also occurred in surgical specimens, we performed real time RT–PCR using 47 RCC tissues and their corresponding normal kidney tissues. Although PPARγ mRNA expression was detectable in both RCC tissues and normal kidney tissues, the expression in RCC tissues relative to normal kidney tissue was variable (Fig. 2). Down-regulation of PPARγ mRNA expression relative to corresponding normal kidney tissues was seen in 25 tissues (tumor : normal ratio corrected by β-actin [T/N]; range, 0.01–0.9). Of these 25 patients, 23 had clear cell carcinoma, 1 spindle cell carcinoma, and 1 Bellini duct carcinoma. PPARγ mRNA expression in RCC tissue was equal to that of normal tissue in one patient. The remaining 21 RCC tissues overexpressed PPARγ mRNA relative to normal kidney tissues (T/N; range, 1.1–8.1).

image

Figure 2. Relative expression of PPARγ mRNA in 47 renal cell carcinoma tissues. Relative expression was determined by the ratio of tumor PPARγ mRNA to corresponding normal kidney tissue PPARγ mRNA, and then was corrected for that with β-actin. ○, patients without metastasis at diagnosis; ●, patients with metastasis at diagnosis.

Download figure to PowerPoint

Relationship between PPARγ mRNA expression and clinicopathological features

Patients were arbitrarily divided into a low expression group (T/N < 1.1) and high expression group (T/N ≥ 1.1). Patients with low expression had distant metastasis at diagnosis more frequently than those with high expression (P = 0.03, Fisher's exact test; Table 2). There was also a trend for patients with low PPARγ expression to have a high stage (P = 0.05, χ2 test). The level of PPARγ mRNA expression in tumors was not correlated with age, gender, grade, or venous invasion.

Table 2.  Clinicopathological features and PPARγ expression in renal cell carcinoma patients
FeaturesLow expression (n = 26) (PPARγ/β-actin < 1.1)High expression (n = 21) (PPARγ/β-actin ≥ 1.1)P-value
  • *

    Mann–Whitney U-test;

  • †Fisher's exact test;

  • χ2-test.

Age (mean)64.8 (Years)62.5 (Years)NS*
Gender
 Male19 (73%)15 (71%)NS
 Female 7 (27%) 6 (29%) 
Pathological stage
 pT1a 4 (15%) 9 (43%) 
 pT1b12 (46%) 6 (29%)0.05
 pT2 1 (4%) 3 (14%) 
 pT3 9 (35%) 3 (14%) 
Grade
 1 2 (8%) 4 (19%) 
 215 (58%)15 (71%)NS
 3 9 (34%) 2 (10%) 
Metastasis at diagnosis
 No18 (69%)20 (95%)0.03
 Yes 8 (31%) 1 (5%) 

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

Several studies have demonstrated that PPARγ expression in cancer tissue is inconstant when compared to normal counterparts. In colon cancer and pancreatic cancer, the expression of PPARγ is equal to or higher in cancer tissue than in corresponding normal tissue.18,19 On the other hand, decreased expression of PPARγ was frequently seen in esophageal, lung, and thyroid cancers.9–11 In this study, we showed that RCC tissues frequently exhibited a lower level of PPARγ expression than normal kidney tissues. Our results are in contrast with the report of Inoue et al.20 who showed that the expression of PPARγ protein was higher in RCC tissue than in normal kidney tissue by immunohistochemical studies. However, our preliminary immunohistochemical studies showed underexpression of PPARγ protein in some RCC tissues that also had lower expression of PPARγ mRNA in real-time RT–PCR (data not shown). In this study we selected real-time RT–PCR for analysis of PPARγ expression, because of the disadvantage that immunohistochemical techniques cannot quantify it correctly. Possible explanations for the discrepancy may be the different PPARγ antibodies and the different tissues used for analysis in each study.

The mechanism by which PPARγ expression is down-regulated in RCC cells remains unclear. It may be related to PPARγ gene mutation. Indeed, Sarraf et al.12 reported functional mutations of the PPARγ gene within exon 3 and exon 5 in 4 of 55 sporadic colon cancers. However, other studies demonstrated that no mutation was detectable in clinical samples and cell lines including colon, prostate, breast, lung, and thyroid cancers, as well as leukemia.11,21 In the current study we could not find any mutations in exon 3 and exon 5 of the PPARγ gene in six human cell lines. No mutations in these exons were identified either in 10 randomly selected RCC tissues or normal tissue (data not shown). Thus, mutation of the PPARγ gene is unlikely to contribute to down-regulation of PPARγ, although we cannot completely exclude the possibility of mutation in other regions of the PPARγ gene. Alternatively, the PPARγ gene is located at chromosome 3p25, in a region where loss of heterozygosity is frequent in RCC.22 If loss of heterozygosity in this region reflects loss of one PPARγ allele, a gene dosage effect could be responsible for underexpression of PPARγ. Further studies on genetic and epigenetic mechanisms such as hypermethylation within the promotor region are necessary to elucidate these questions.

What are the clinical implications of down-regulation of PPARγ expression? Recent studies have demonstrated that a lower level of PPARγ expression is correlated with tumor invasiveness, and furthermore, worse prognoses in lung, esophageal, and breast cancers.9,10,23 More recently, Marques et al.24 also reported that follicular thyroid cancer with negative PPARγ expression tended to be locally invasive, and metastasize more frequently. Similarly, we found that at diagnosis RCC patients with low PPARγ expression had distant metastasis more frequently, and seemingly a higher stage. Thus, down-regulation of PPARγ expression seems to occur in cancer with an aggressive nature. It is poorly understood why decreased PPARγ is related to the tumor aggressiveness. PPARγ is known to modulate metastasis-related genes such as matrix metalloproteinase-2 (MMP-2) and E-cadherin, and angiogeneic factors such as vascular endothelial growth factor and basic fibroblast growth factor.8,25,26 One possible explanation may be up-regulation of MMP-2 and angiogenic factors, and down-regulation of E-cadherin, which favor invasion and metastasis.

When we consider the therapeutic use of PPARγ ligands in RCC patients, we may be concerned about resistance to this treatment in patients with low levels of PPARγ expression. However, our in vitro experiments using human RCC cell lines showed that there was no correlation between inhibitory effects of PPARγ ligands and PPARγ expression levels. These findings suggest that, regardless of the level of PPARγ expression, PPARγ ligands may be worth trying for treatment of RCC patients, although at present we cannot predict the effect of the ligands with certainty.

In conclusion, we showed that a significant proportion of renal cell carcinomas as well as RCC cell lines had down-regulated expression of PPARγ mRNA. These carcinomas might have more aggressive behavior. In addition to being a potential novel therapeutic target in RCC patients as suggested in a previous study,8 the current findings suggest that PPARγ mRNA expression might be a prognostic marker. Further large-scale studies will be needed to verify this hypothesis.

Acknowledgments

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

We thank Ms Toshie Kurohata for technical support. This work was supported in part by a Grant-in-Aid from the Japanese Ministry of Education, Science, Sports and Culture, and the Stiftelsen Japanese–Swedish Cooperative Research Foundation.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  • 1
    Mangelsdorf DJ, Thummel C, Beato M et al. The nuclear receptor superfamily: the second decade. Cell 1995; 83: 8359.
  • 2
    Desvergne B, Wahli W. Peroxisome proliferator-activated receptors: nuclear control of metabolism. Endocr. Rev. 1999; 20: 64988.
  • 3
    Jiang C, Ting AT, Seed B. PPARγ agonists inhibit production of monocyte inflammatory cytokines. Nature 1998; 391: 826.
  • 4
    Tontonoz P, Nagy L, Alvarez JG, Thomazy VA, Evans RM. PPARγ promotes monocyte/macrophage differentiation and uptake of oxidized LDL. Cell 1998; 93: 24152.
  • 5
    Spiegelman BM. PPARγ in monocytes: less pain, any gain? Cell 1998; 93: 1535.
  • 6
    Koeffler HP. Peroxisome proliferator-activated receptor γ and cancers. Clin. Cancer Res. 2003; 9: 19.
  • 7
    Grommes C, Landreth GE, Heneka MT. Antineoplastic effects of peroxisome proliferator-activated receptor gamma agonists. Lancet Oncol. 2004; 5: 41929.
  • 8
    Yuan J, Takahashi A, Masumori N et al. Ligands for peroxisome proliferator-Activated receptor gamma have a potent antitumor effect against human renal cell carcinoma. Urology 2005; 65: 5949.
  • 9
    Terashita Y, Sasaki H, Haruki N et al. Decreased peroxisome proliferator-activated receptor gamma gene expression is correlated with poor prognosis in patients with esophageal cancer. Jpn. J. Clin. Oncol. 2002; 32: 23843.
  • 10
    Sasaki H, Tanahashi M, Yukiue H et al. Decreased perioxisome proliferator-activated receptor gamma gene expression was correlated with poor prognosis in patients with lung cancer. Lung Cancer 2002; 36: 716.
  • 11
    Aldred MA, Morrison C, Gimm O et al. Peroxisome proliferator-activated receptor γ is frequently downregulated in a diversity of sporadic nonmedullary thyroid carcinomas. Oncogene 2003; 22: 341216.
  • 12
    Sarraf P, Mueller E, Smith WM et al. Loss-of-function mutations in PPARγ associated with human colon cancer. Mol. Cell 1999; 3: 799804.
  • 13
    Miyao N, Tsukamoto T, Kumamoto Y. Establishment of three human renal cell carcinoma cell lines (SMKT-R-1, SMKT-R-2, and SMKT-R-3) and their characters. Urol. Res. 1989; 17: 31724.
  • 14
    Otani N, Tsukamoto T, Kumamoto Y, Miyao N. Study on in vivo invasive potential of renal cell carcinoma cell lines and effect of growth factors (EGF and TGF-β1) on their in vivo invasions. Jpn. J. Urol. 1991; 82: 61319.
  • 15
    Sobin LH, Wittekind C. TNM Classification of Malignant Tumours, 6th edn. Wiley-Liss, New York, 2002; 1935.
  • 16
    Pavan L, Tarrade A, Hermouet A et al. Human invasive trophoblasts transformed with simian virus 40 provide a new tool to study the role of PPARγ in cell invasion process. Carcinogenesis 2003; 24: 132536.
  • 17
    Haslmayer P, Thalhammer T, Jager W et al. The peroxisome proliferator-activated receptor γ ligand 15-deoxy-Δ12,14-prostaglandin J2 induces vascular endothelial growth factor in the hormone-independent prostate cancer cell line PC 3 and the urinary bladder carcinoma cell line 5637. Int. J. Oncol. 2002; 21: 91520.
  • 18
    DuBois RN, Gupta R, Brockman J, Reddy BS, Krakow SL, Lazar MA. The nuclear eicosanoid receptor, PPARγ, is aberrantly expressed in colonic cancers. Carcinogenesis 1998; 19: 4953.
  • 19
    Motomura W, Okumura T, Takahashi N, Obara T, Kohgo Y. Activation of peroxisome proliferator-activated receptor γ by troglitazone inhibits cell growth through the increase of p27KiP1 in human pancreatic carcinoma cells. Cancer Res. 2000; 60: 555864.
  • 20
    Inoue K, Kawahito Y, Tsubouchi Y et al. Expression of peroxisome proliferator-activated receptor γ in renal cell carcinoma and growth inhibition by its agonists. Biochem. Biophys. Res. Commun. 2001; 287: 72732.
  • 21
    Ikezoe T, Miller CW, Kawano S et al. Mutational analysis of the peroxisome proliferator-activated receptor γ gene in human malignancies. Cancer Res. 2001; 61: 530710.
  • 22
    Shuin T, Kondo K, Torigoe S et al. Frequent somatic mutations and loss of heterozygosity of the von Hippel-Lindau tumor suppressor gene in primary human renal cell carcinomas. Cancer Res. 1994; 54: 28525.
  • 23
    Jiang WG, Douglas-Jones A, Mansel RE. Expression of peroxisome-proliferator activated receptor-gamma (PPARγ) and the PPARγ co-activator, PGC-1, in human breast cancer correlates with clinical outcomes. Int. J. Cancer 2003; 106: 7527.
  • 24
    Marques AR, Espadinha C, Frias MJ et al. Underexpression of peroxisome proliferator-activated receptor (PPAR) γ in PAX8/PPARγ-negative thyroid tumours. Br. J. Cancer 2004; 91: 7328.
  • 25
    Stetler-Stevenson WG, Aznavoorian S, Liotta LA. Tumor cell interactions with the extracellular matrix during invasion and metastasis. Annu. Rev. Cell Biol. 1993; 9: 54173.
  • 26
    Ohta T, Elnemr A, Yamamoto M et al. Thiazolidinedione, a peroxisome proliferator-activated receptor-gamma ligand, modulates the E-cadherin/beta-catenin system in a human pancreatic cancer cell line, BxPC-3. Int. J. Oncol. 2002; 21: 3742.