Skip Navigation



NDT Advance Access published online on September 11, 2008
Nephrology Dialysis Transplantation, doi:10.1093/ndt/gfn505
This Article
Right arrow Abstract Freely available
Right arrow FREE Full Text (PDF) Freely available
Right arrow All Versions of this Article:
24/2/470    most recent
gfn505v1
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Archive
Right arrow Download to citation manager
Right arrowRequest Permissions
Right arrow Disclaimer
Google Scholar
Right arrow Articles by Sharma, A. P.
Right arrow Articles by Filler, G.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Sharma, A. P.
Right arrow Articles by Filler, G.
Social Bookmarking
Add to CiteULike   Add to Connotea   Add to Del.icio.us  
What's this?

© The Author [2008]. Published by Oxford University Press on behalf of ERA-EDTA. All rights reserved. For Permissions, please e-mail: journals.permissions@oxfordjournals.org


Body mass does not have a clinically relevant effect on Cystatin C eGFR in children

Ajay P. Sharma1, Anusha Kathiravelu2, Renisha Nadarajah2, Abeer Yasin1 and Guido Filler1

1 Department of Paediatrics, Division of Paediatric Nephrology, Children's Hospital, London Health Science Centre, University of Western Ontario, London 2 Division of Pediatric Nephrology, Children's Hospital of Eastern Ontario, University of Ottawa, Ottawa, Ontario, Canada

Correspondence and offprint requests to: Guido Filler, Children's Hospital, London Health Science Centre, University of Western Ontario, 800 Commissioner's Road East, London, Ontario, N6A 5W9, Canada. Tel: +1-519-685-8377; Fax: +1-519-685-8551; E-mail: guido.filler{at}lhsc.on.ca



   Abstract
Top
 Abstract
Introduction
 Subjects and methods
 Results
 Discussion
 References
 
Background. Unlike creatinine, Cystatin C (CysC) is believed to be independent of body composition in both adults and children. Recent findings in adults, suggesting an improved performance of CysC-based estimated glomerular filtration rate (CysC eGFR) by accounting for body mass, necessitated a careful re-evaluation of this issue in children.

Methods. We studied 240 children (median age 11.7 years, range 2–17.9 years, 107 girls), with various kidney diseases, for any change in the relationship between 99Tc DTPA GFR and CysC eGFR after accounting for body mass. For body mass assessment, body mass index (BMI) z-score was calculated using height-adjusted age, to account for growth retardation secondary to chronic kidney disease.

Results. CysC eGFR did not have a significant correlation with BMI z-score (correlation coefficient = 0.06; P = 0.34). Accounting for BMI z-score did not add to the 65% variance in nuclear GFR explained by CysC eGFR. Moreover, it did not change the regression coefficient of 0.85 between CysC eGFR and nuclear GFR either. On Bland & Altman analysis, the bias of 0.05 and standard deviation of 20.39 also did not improve after accounting for BMI z-score in the revised CysC eGFR formula.

Conclusions. In children, body mass exerts a minimal effect on the performance of CysC eGFR estimation.

Keywords: body composition; body mass index; Cystatin C; obesity; z-score



   Introduction
Top
 Abstract
 Introduction
Subjects and methods
 Results
 Discussion
 References
 
Glomerular filtration rate (GFR) determination is essential for the management of patients with kidney disease. For a direct GFR estimation, inulin clearance is the gold-standard method; however, it is expensive, cumbersome and is not readily available at many centres. As a result, nuclear medicine techniques have largely replaced inulin clearance in routine clinical practice. Despite the advantage of a direct GFR assessment, radiation exposure, associated cost and multiple blood samples limit a frequent use of nuclear medicine techniques in an individual patient [1,2]. Consequently, frequent kidney function assessments largely rely on estimated GFR (eGFR) derived from endogenous surrogate markers.

Conventionally, creatinine has been the most frequently used surrogate marker for eGFR estimation. Recently, Cystatin C (CysC) has been found to have a 10% better diagnostic performance than serum creatinine in mild renal impairment [3]. Regardless of the marker or the formula used to calculate eGFR, a 30–40% scatter between measured GFR and eGFR remains unexplained. Imprecision of the analytical methods can account for only 10% of the variance [4]. Although the recognized effect of muscle mass on serum creatinine-based eGFR can further explain a part of the scatter, a similar confounding effect of body mass on CysC eGFR is not clear. The body mass and CysC relationship has been the focus of a few recent adult and paediatric studies.

Earlier, Vinge et al. demonstrated no correlation between CysC and lean tissue mass in 42 healthy young adults with normal renal functions [5]. Similarly, Bökenkamp et al. did not find any correlation of CysC with weight or height in children with chronic kidney disease (CKD) [6]. In contrast, Galteau et al. reported a moderate correlation between body mass index (BMI) and CysC in older subjects, but no such association existed in the enrolled children [7]. This large study on 1223 subjects raised a possibility of different relationship between body composition and CysC eGFR in adults and children.

To expand on this debate, MacDonald et al. reported 6% variance in CysC by lean mass, and 16% improved performance by accounting for body mass in the CysC eGFR formula, despite no direct correlation between body mass and CysC similar to the earlier studies [8]. This study involved adult CKD patients. We are not aware of any similar paediatric study that assessed the effect of body mass on CysC variance, and evaluated the performance of CysC eGFR estimation after accounting for body mass. Based on the fact that CysC is generated from all the nucleated cells in the body [9,10], we hypothesized a similar effect of body mass on CysC in children too, although the degree of effect may be different in children due to their smaller body sizes.

To test this hypothesis, we assessed CysC variance secondary to body mass in paediatric CKD patients. In addition, we analysed any improvement in the correlation between nuclear GFR and CysC eGFR by accounting for body mass.



   Subjects and methods
Top
 Abstract
 Introduction
 Subjects and methods
Results
 Discussion
 References
 
The study was approved by the Institutional Review Board approval. We performed a post hoc analysis of the prospectively collected data on nuclear GFR and CysC levels from 240 children referred to the paediatric nephrology clinic. These measurements occurred within a larger study of 536 children enrolled to derive the CysC formula [4]. The data on age, weight, and height were collected to calculate body surface area (BSA), BMI and BMI z-scores [11]. Patients with nuclear GFR estimated by 51CR EDTA, or with incomplete anthropometric measurements, were excluded. Patients aged 2 years or younger were also excluded due to unavailability of reference data for BMI z-score calculation.

Nuclear GFR was measured using a 99m technetium-diethylene-triamine penta-acetic acid (99mTc DTPA) renal scan with a three-point sampling approach at 2, 3 and 4 h post-injection according to Russell [12]. As per conventional practice, the measured 99mTc DTPA GFR was normalized to BSA of 1.73 m2, to account for the change in body size with age. BSA and BMI were calculated by the Haycock formula [13], and by the ratio of weight (kg) and square of height (m), respectively. To calculate BMI z-scores, age and gender-specific standard deviations (SDs) published by the National Center for Health Statistics [14] were used. The method to test CysC (nephelometry, Dade Behring kit) has been described elsewhere [15]. CysC eGFR was determined using the previously published formula, using regression analysis of the log–log transformed data [15]. It reads CysC eGFR = 10^(1.962 + 1.123 log(1/CysC)) (mL/min/1.73 m2) (formula 1). The old formula was updated for this specific data set with the changed constants accordingly. The formula now reads CysC eGFR = 10^(1.948 + 1.184 log(1/CysC)); R = 0.921, R2 = 0.848 (formula 2).

With the CysC eGFR derived from formula 2, we built the following regression models to analyse the relationship between nuclear GFR and CysC eGFR for the entire group, and also separately in females and males, with and without accounting for BMI z-score. Notably, BMI z-score estimation was done for both chronological age (model 2) and height-adjusted age (model 3). The use of height-adjusted age was intended to correct for a detrimental effect of CKD on height, as earlier validated by Schaefer et al. [16].

Entire group

  • Model 1: GFR = 19.134 + 0.819 x CysC eGFR (formula 2); R = 0.808, R2 = 0.653.
  • Model 2: GFR = 19.229 + 0.807 x CysC GFR (formula 2) + 3.369 x BMI z-score; R = 0.814, R2 = 0.662.
  • Model 3: GFR = 18.635 + 0.816 x CysC eGFR (formula 2) + 2.007 x BMI z-score using height-adjusted age; R = 0.810, R2 = 0.656.

Females

  • Model 1: GFR = 30.550 + 0.700 x CysC eGFR (formula 2); R = 0.720, R2 = 0.519.
  • Model 2: GFR = 29.461 + 0.699 x CysC eGFR (formula 2) + 2.456 x BMI z-score; R = 0.723, R2 = 0.523.
  • Model 3: GFR = 29.414 + 0.700 x CysC eGFR (formula 2) + 2.064 x BMI z-score using height-adjusted age; R = 0.722, R2 = 0.522.

Males

  • Model 1: GFR = 9.026 + 0.930 x CysC eGFR (formula 2); R = 0.882, R2 = 0.777.
  • Model 2: GFR = 10.331 + 0.910 x CysC eGFR (formula 2) + 3.662 x BMI z-score; R = 0.888, R2 = 0.789.
  • Model 3: GFR = 8.914 + 0.926 x CysC eGFR (formula 2) + 1.967 x BMI z-score using height-adjusted age; R = 0.883, R2 = 0.781.

To further analyse the effect of BMI z-score on CysC eGFR, we derived a revised formula for CysC eGFR estimation that incorporated BMI z-score. The revised formula reads CysC eGFR = 10^(1.944 + 1.181 log(1/CysC) + 0.009 x BMI z-score using height-adjusted age (HAA)); R = 0.922, R2 = 0.850 (formula 3). To account for gender-specific differences, separate formulae were derived for the boys and girls. In boys, without accounting for BMI z-score, it reads GFR = 10^(1.956 + 1.221 log(1/CysC)), R = 0.941, R2 = 0.885; and after accounting for BMI z-score, it reads GFR = 10^(1.952 + 1.217 log(1/CysC) + 0.013 BMI Z-HAA), R = 0.943, R2 = 0.890. In girls, the corresponding formulae read GFR = 10^(1.938 + 1.149 log(1/CysC)), R = 0.898, R2 = 0.807; and GFR = 10^(1.935 + 1.149 log(1/CysC) + 0.005 x BMI Z-HAA), R = 0.899, R2 = 0.808.

For a sub-group analysis in obese children, we derived similar formulae in the children with BMI z-score >2.0 SDs. These formulae read CysC eGFR = 10^(1.962+1.123 log(1/CysC), R = 0.913, R2 = 0.833, without accounting for BMI z-score; and CysC eGFR = 10^(1.962 + 1.123 log(1/CysC) +1.06x10–6 log BMI Z-HAA), R = 0.914, R2 = 0.835, after accounting for BMI z-score.

Statistical analysis
Wherever possible, simple descriptive statistics were used. Contiguous data were tested for normal distribution using the D’Agostini Pearson omnibus test and the Shapiro–Wilks test. Normally distributed data were analysed using parametric methods (mean, SD, t-test, Pearson correlation); otherwise non-parametric methods were employed (median, range, Wilcoxon's matched pairs test and Spearman rank correlation). All statistical analyses were conducted using GraphPad Prism software, version 4.02 (GraphPad Inc., San Diego, CA, USA). For the multiple regression analyses, SPSS version 15 (SPSS Inc., Chicago, IL, USA) was also utilized.



   Results
Top
 Abstract
 Introduction
 Subjects and methods
 Results
Discussion
 References
 
The study group constituted 240 patients. Patient characteristics are summarized in Table 1. Based on the BMI estimation according to the chronological age, BMI z-scores in excess of 2.0 (obese group) were present in 17 (7%) children, whereas corresponding to the height-adjusted age this proportion was 19 (8%).


View this table:
[in this window]
[in a new window]

  		Table 1 Patient characteristics, GFR estimates and anthropometric measurements

 
The correlation between nuclear GFR with CysC eGFR, and also with other variables, is summarized in Table 2. As expected, nuclear GFR significantly correlated with CysC eGFR [correlation coefficient (c) = 0.83; P < 0.000] (Figure 1). On the other hand, nuclear GFR did not have a significant correlation with age, weight, height, BMI and BMI z-score in the whole group or in the separate analyses conducted on both genders and obese children.


View this table:
[in this window]
[in a new window]

  		Table 2 Correlations between CysC eGFR with other variables

 

Figure 1
View larger version (16K):
[in this window]
[in a new window]
[Download PowerPoint slide]
  		Fig. 1 The correlation coefficient between nuclear GFR (ml/min/ 1.73 m2) and the estimated Cystatin C GFR (ml/min/1.73 m2) after accounting for BMI z-score in the formula for Cystatin C GFR.

 
On regression analysis, the regression coefficient between nuclear GFR and CysC eGFR, without accounting for BMI z-score, was 0.80 in the whole group (Table 3), improved to 0.88 in a separate analysis on the boys, and only reached 0.72 in the girls. After accounting for BMI z-score, the regression coefficient improved only marginally to 0.85 in the whole group, dropped minimally to 0.87 in the boys and remained at 0.72 in the girls. In the obese group, the regression coefficient marginally changed from 0.91 to 0.92 after accounting for BMI z-score. Furthermore, without accounting for BMI z-score, CysC eGFR explained 65% variance in nuclear GFR in the whole group, 77% in the boys and 51% in the girls. After accounting for BMI z-score, the variance explained by CysC eGFR largely remained unchanged at 65% in the whole group, 78% in boys and 52% in the girls (Table 3). In the obese group, CysC eGFR explained 83% variance in nuclear GFR, and it remained unchanged after accounting for BMI z-score.


View this table:
[in this window]
[in a new window]

  		Table 3 Regression models between nuclear GFR and Cystatin C eGFR with and without accounting for BMI z-score

 
To evaluate the performance of the CysC eGFR formula that accounted for body mass (formula 3), we compared its agreement with nuclear GFR as compared with the formula that did not account for body mass (formula 2). On Bland & Altman analysis, without accounting for BMI z-score, the respective bias and SD amounted to +0.05 and 20.39 in the whole group (Figure 2), –0.077 and 17.96 in the boys and +0.05 and 22.77 in the girls (Table 4). After accounting for BMI z-score in the revised formula, the bias and SD in the whole group turned out to be –0.05 and 20.23, respectively (Figure 3), with the corresponding values of –0.03 and 22.74 in the boys and –0.03 and 22.74 in the girls. In the obese group also, there was no change in the bias and SD from the respective values of –2.00 and 15.91 after accounting for BMI z-score.


Figure 2
View larger version (18K):
[in this window]
[in a new window]
[Download PowerPoint slide]
  		Fig. 2 Bland & Altman analysis: the agreement between nuclear GFR (ml/min/1.73 m2) and estimated Cystatin C GFR (ml/min/1.73 m2) without accounting for BMI z- score in the formula used for Cystatin C eGFR (bias 0.05; standard deviation 20.39).

 

View this table:
[in this window]
[in a new window]

  		Table 4 Bland & Altman analysis for the agreement of two CysC eGFR formulae (with and without accounting for body mass) with nuclear GFR

 

Figure 3
View larger version (17K):
[in this window]
[in a new window]
[Download PowerPoint slide]
  		Fig. 3 Bland & Altman analysis: the agreement between nuclear GFR (ml/min/1.73 m2) and the estimated Cystatin C GFR (ml/min/1.73 m2) after accounting for BMI z-score in the revised formula for Cystatin C eGFR (bias –0.05; standard deviation 20.23).

 


   Discussion
Top
 Abstract
 Introduction
 Subjects and methods
 Results
 Discussion
References
 
A lack of statistically significant correlation between CysC and body mass in the previous studies on both adults and children led to the earlier belief that CysC-based eGFR is independent of body composition [5–7]. The validity of this relationship was questioned by a recent study in the adult CKD patients reporting a variance in CysC levels secondary to body mass, and an improved performance of the CysC eGFR formula after accounting for body mass [8]. Based on this new observation, we revisited the relationship between body mass and CysC eGFR in children.

As a first step, we evaluated the correlation between CysC eGFR and BMI z-score. Importantly, to account for age dependence of BMI in children, we calculated the corresponding BMI z-scores. The association between BMI z-score and CysC eGFR did not turn out to be significant either in the whole study sample or in the separate analyses on the boys, girls and obese children. This observation was similar to the lack of correlation between body mass and CysC eGFR in the previous studies in both adults and children [5–7].

As the next step, we analysed CysC variance secondary to body mass, and change in the performance of the CysC eGFR formula by accounting for body mass. Without accounting for BMI z-score, the CysC eGFR formula explained 65% variance in nuclear GFR in the whole group, 77% in the boys, 51% in the girls and 83% in the obese group. Accounting for BMI z-score did not change the variance either in the whole group or in the separate subgroup analyses in boys, girls and obese children. This observation was different from a 16% extra variance explained by the revised formula that accounted for body mass in the adult subjects evaluated by MacDonald et al. [8]. Furthermore, accounting for BMI z-score did not significantly affect the regression coefficient between nuclear GFR and CysC eGFR in our patients as a whole or in the separate subgroups.

We also analysed the agreement of nuclear GFR with the two CysC eGFR formulae, one that did not account for BMI z-score (formula 2) and the other that accounted for BMI z-score (formula 3). On Bland & Altman analysis, the bias of 0.05 and SD of 20.39 with formula 2 did not significantly change after accounting for BMI z-score in formula 3. The separate analyses in both genders, and also in the obese children, did not reveal any improvement with formula 3 either. These results were again different from the improved agreement after accounting for body mass in the CysC eGFR formula in the adult CKD patients [8]. Unlike the adult patients [8], a weak effect of body mass on the CysC eGFR and nuclear GFR relationship could be a reflection of overall lower CysC generation in children, owing to a smaller nucleated cell pool corresponding to a relatively smaller body mass [9].

Unlike the use of BMI [7], weight and height [6] in the earlier studies for body mass estimation, we calculated BMI z-scores for this purpose. Although BMI estimation corrects for a variation in weight for the corresponding height, it does not account for an age-dependent increase in anthropometric parameters in children. On the other hand, BMI z-score offers the advantage of adjusting for this confounder. Furthermore, CKD can also confound BMI assessment in children as a result of its detrimental effect on growth velocity. To address this issue, we calculated BMI z-score based on height-adjusted age, rather than using the corresponding values for the chronological age [16,17].

Our study has a few limitations. Due to an exclusive Caucasian study sample, our observations are not applicable to CKD patients from other ethnicities. To assess body mass, the use of BMI z-score, instead of Dual Energy X-ray Absorptiometry (DEXA), can be argued. It is important to note that the presence of CKD poses an inherent challenge in accurate body mass assessment due to fluid overload and the abnormalities in the distribution of fat and lean tissue [17,18]. In children with CKD, the same set of challenges exists with DEXA too, and also the available data with DEXA have not been evaluated after adjustment for height [17]. Although a component of similar limitations cannot be completely excluded with BMI; nevertheless, BMI corresponding for height-age has been suggested as a reasonably appropriate method of standardization in paediatric CKD patients [16,17]. Incidentally, at the higher GFR range, there was a relative widening of the scatter between nuclear GFR and CysC eGFR in our subjects. Although unexplained, we postulate this widening at high GFR due to different clearance characteristics estimated by 99Tc DTPA and CysC, predominantly water clearance by a small molecule like 99Tc DTPA and middle molecule clearance by CysC.

We conclude that accounting for body mass does not improve the performance of CysC eGFR estimate in paediatric CKD patients. This issue needs further validation in larger studies.



   Acknowledgments
 
G.F. received research grants (operating grants and contributions to run free assays for the measurement of CysC) from Dade Behring GmbH, Marburg, Germany. No honoraria were paid.

Conflict of interest statement. None declared.



   References
Top
 Abstract
 Introduction
 Subjects and methods
 Results
 Discussion
 References
 

  1. Pabst HW. Nuclear medicine clearance procedures in nephrology. Fortschr Geb Rontgenstr Nuklearmed (1972) (Suppl):94–95.
  2. Krawiec DR, Badertscher RR II, Twardock AR, et al. Evaluation of 99mTc-diethylenetriaminepentaacetic acid nuclear imaging for quantitative determination of the glomerular filtration rate of dogs. Am J Vet Res (1986) 47:2175–2179.[Web of Science][Medline]
  3. Dharnidharka VR, Kwon C, Stevens G. Serum cystatin C is superior to serum creatinine as a marker of kidney function: a meta-analysis. Am J Kidney Dis (2002) 40:221–226.[CrossRef][Web of Science][Medline]
  4. Filler G, Priem F, Lepage N, et al. Beta-trace protein, cystatin C, beta(2)-microglobulin, and creatinine compared for detecting impaired glomerular filtration rates in children. Clin Chem (2002) 48:729–736.[Abstract/Free Full Text]
  5. Vinge E, Lindergard B, Nilsson-Ehle P, et al. Relationships among serum cystatin C, serum creatinine, lean tissue mass and glomerular filtration rate in healthy adults. Scand J Clin Lab Invest (1999) 59:587–592.[CrossRef][Web of Science][Medline]
  6. Bokenkamp A, Domanetzki M, Zinck R, et al. Cystatin C—a new marker of glomerular filtration rate in children independent of age and height. Pediatrics (1998) 101:875–881.[Abstract/Free Full Text]
  7. Galteau MM, Guyon M, Gueguen R, et al. Determination of serum cystatin C: biological variation and reference values. Clin Chem Lab Med (2001) 39:850–857.[CrossRef][Web of Science][Medline]
  8. Macdonald J, Marcora S, Jibani M, et al. GFR estimation using cystatin C is not independent of body composition. Am J Kidney Dis (2006) 48:712–719.[CrossRef][Medline]
  9. Abrahamson M, Olafsson I, Palsdottir A, et al. Structure and expression of the human cystatin C gene. Biochem J (1990) 268:287–294.[Web of Science][Medline]
  10. Maack T, Johnson V, Kau ST, et al. Renal filtration, transport, and metabolism of low-molecular-weight proteins: a review. Kidney Int (1979) 16:251–270.[Web of Science][Medline]
  11. Filler G, Payne RP, Orrbine E, et al. Changing trends in the referral patterns of pediatric nephrology patients. Pediatr Nephrol (2005) 20:603–608.[CrossRef][Web of Science][Medline]
  12. Russell CD. Optimum sample times for single-injection, multisample renal clearance methods. J Nucl Med (1993) 34:1761–1765.[Abstract/Free Full Text]
  13. Haycock GB, Schwartz GJ, Wisotsky DH. Geometric method for measuring body surface area: a height-weight formula validated in infants, children, and adults. J Pediatr (1978) 93:62–66.[Web of Science][Medline]
  14. http://www.cdc.gov/nchs/about/major/nhanes/growthcharts/zscore/zscore.htm (last accessed 23 June 2006).
  15. Filler G, Lepage N. Should the Schwartz formula for estimation of GFR be replaced by cystatin C formula? Pediatr Nephrol (2003) 18:981–985.[CrossRef][Web of Science][Medline]
  16. Schaefer F, Wuhl E, Feneberg R, et al. Assessment of body composition in children with chronic renal failure. Pediatr Nephrol (2000) 14:673–678.[CrossRef][Web of Science][Medline]
  17. Foster BJ, Leonard MB. Measuring nutritional status in children with chronic kidney disease. Am J Clin Nutr (2004) 80:801–814.[Abstract/Free Full Text]
  18. Raffaitin C, Lasseur C, Chauveau P, et al. Nutritional status in patients with diabetes and chronic kidney disease: a prospective study. Am J Clin Nutr (2007) 85:96–101.[Abstract/Free Full Text]
Received for publication: 11. 4.08
Accepted in revised form: 18. 8.08


Add to CiteULike CiteULike   Add to Connotea Connotea   Add to Del.icio.us Del.icio.us    What's this?



This Article
Right arrow Abstract Freely available
Right arrow FREE Full Text (PDF) Freely available
Right arrow All Versions of this Article:
24/2/470    most recent
gfn505v1
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Archive
Right arrow Download to citation manager
Right arrowRequest Permissions
Right arrow Disclaimer
Google Scholar
Right arrow Articles by Sharma, A. P.
Right arrow Articles by Filler, G.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Sharma, A. P.
Right arrow Articles by Filler, G.
Social Bookmarking
Add to CiteULike   Add to Connotea   Add to Del.icio.us  
What's this?