Letter to the Editor
Response to letter to the editor ‘Serum cystatin-C levels correlate with endothelial dysfunction in patients with the metabolic syndrome’
Article first published online: 22 JUL 2013
© 2013 The Association for the Publication of the Journal of Internal Medicine
Journal of Internal Medicine
Volume 274, Issue 5, pages 496–498, November 2013
How to Cite
Magnusson, M., Hedblad, B., Engström, G., Persson, M., Nilsson, P. and Melander, O. (2013), Response to letter to the editor ‘Serum cystatin-C levels correlate with endothelial dysfunction in patients with the metabolic syndrome’. Journal of Internal Medicine, 274: 496–498. doi: 10.1111/joim.12110
- Issue published online: 16 OCT 2013
- Article first published online: 22 JUL 2013
- Accepted manuscript online: 10 JUL 2013 06:44AM EST
- Manuscript Accepted: 4 JUN 2013
We thank the authors for their interest in our article entitled ‘High levels of cystatin C predict the metabolic syndrome: the prospective Malmo Diet and Cancer Study’ .
In the letter to the editor titled ‘Serum cystatin-C levels correlate with endothelial dysfunction in patients with the metabolic syndrome’ , Dr. Balta and co-workers propose that there may be an indirect relationship between serum cystatin C (s-CC) levels and CVD risk through renal dysfunction, and its effect on endothelial function and the general message made by Dr. Balta and his co-workers, as we understand it, is the consideration of additional cofactors (not addressed in our article) that might affect kidney function (e.g. as measured by s-CC) and thereby influence the prognostic value of serum cystatin C (s-CC) regarding future development of the metabolic syndrome (MetS) found in our study . In our response, we will try to address the authors’ comments point by point.
As a general remark, we would like to emphasize that to study the specific effects of the CC-metabolism (not biased by declining renal clearance) on metabolic risk factors, we excluded all patients with known renal dysfunction at baseline examination from the study (renal dysfunction was defined as glomerular filtration rate (GFR) below 60 mL−1 min/1.73 m2). Furthermore, because renal dysfunction at a GFR above 60 mL−1 min/1.73 m2 might affect s-CC levels, we reanalysed the association between s-CC and incident MetS including GFR as a cofactor in addition to the others in the full multivariate logistic regression analysis. This analysis showed that each 1 SD increment in baseline levels of s-CC was still highly significantly associated with an approximately 20 per cent increased risk of incident MetS (OR 1.21, 95% CI 1.06–1.39, P = 0.006).
Because the elevation of inflammatory mediators is a common sign of atherosclerotic involvement in the vascular structure, indicating the presence of coronary artery disease, cerebrovascular disease, peripheral arterial disease or chronic inflammation, Dr. Balta et al. find it noteworthy that alcohol consumption, insulin resistance, hypothyroidism, heart failure, cerebrovascular disease, peripheral arterial disease and chronic inflammatory disease were not assessed in this study.
To address these interesting points, we reperformed our statistical analysis regarding baseline values of s-CC and incident MetS as stated below.
None of 1502 patients without MetS at baseline examination had a history of heart failure. 1.5 per cent of the study population (n = 23) had a history of ischaemic heart disease, and only 3 subjects had a history of cerebrovascular disease. Given the low prevalence of known cardiovascular disease (CVD) at baseline examination, when entering CVD as a cofactor in the multivariate analysis, it was not surprising that the results were not altered in any significant way, and each 1 SD increment in baseline levels of cystatin C still significantly increased the risk of incident MetS (OR 1.15, 95% CI 1.00–1.31, P = 0.046). Regarding alcohol consumption, we have data for 1480 of the 1502 subjects (mean value 10.6 gram consumed/day), and when added to the other covariates in the full multivariate logistic regression analysis model, alcohol consumption itself showed no significant correlation with incident MetS (OR 1.00, 95% CI 0.99–1.02, P = 0.570), and s-CC association with MetS was only slightly weakened (OR 1.14, 95% CI 1.00–1.31, P = 0.055) most likely due to a loss of power because of 22 subjects with missing values. Finally, regarding insulin resistance, we used the homoeostatic model assessment of insulin resistance (HOMA-IR)  to estimate insulin resistance and divided the population into quartiles and defined insulin resistance as being part of the top quartile of the study population (which now was decreased to 1400 subjects due to missing values of s-insulin). Insulin resistance was, as expected, significantly associated with incident MetS when entered on top of all other variables (glucose excluded because it is included in HOMA-IR) in the multivariate logistic regression analysis (OR 1.66, 95% CI 1.19–2.30, P = 0.003), and s-CC association with incident MetS was significantly weakened (OR 1.11, 95% CI 0.96–1.27, P = 0.153). Unfortunately, because we lack data regarding hypothyroidism and/or hyperthyroidism, we cannot account for possible effects of the thyroid hormones that might have had on the CC-metabolism.
The authors also ask for the consideration of antihypertensive treatment (including angiotensin-converting-enzyme inhibitors, angiotensin receptor blockers and beta blockers), statins and medication for weight-loss effects on S-CC levels. Indeed, in the original article, antihypertensive treatment (comprising angiotensin-converting-enzyme inhibitors, angiotensin receptor blockers and beta blockers) was included in the multivariate logistic regression analysis regarding all endpoints, including MetS, which is clearly stated throughout the manuscript . However, because lipid-lowering medication was not included in the multivariate analysis in the original manuscript, we have reperformed the logistic regression for incident MetS, including lipid-lowering medication at baseline examination as a cofactor on top of all the others in the multivariate analysis. 2.2 per cent (n = 33) of the 1502 subjects had lipid-lowering medication at baseline examination, and lipid-powering medication at baseline examination was not statistically significantly associated with incident MetS (OR 1.17, 95% CI 0.51–2.68, P = 0.702, whereas s-CC remained significantly associated with incident MetS in this model (OR 1.15, 95% CI 1.00–1.31, P = 0.044). Unfortunately, we lack data regarding medication for weight loss; however, this would unlikely have influenced our results, because the study population was recruited in the early 1990s (there was to our knowledge no specific medication for weight loss available on the Swedish market at that time).
Finally, Balta and co-workers state that because s-CC has been used as a measure of inflammatory status, s-CC alone without other inflammatory markers may not provide clinicians with sufficient information about the MetS. For this reason, they propose that s-CC levels should be evaluated together with other serum inflammatory markers in routine clinical practice. This is an interesting point made by the authors. We therefore entered high-sensitive CRP (which was available in all 1502 subjects) on top of all the other covariates in the multiple logistic regression model for incident MetS. Interestingly, CRP showed no significant association with incident MetS (OR 1.19, 95% CI 0.85–1.67, P value = 0.306); however, s-CC remained statistically significantly associated with incident MetS (OR 1.15, 95% CI 1.00–1.31, P = 0.044).
So in summary, the only additional cofactor that weakened s-CC association with incident MetS was insulin resistance, although we cannot rule out that this might have been due to a loss of power because as many as 102 subjects had missing values of serum insulin.
However, given the strong association between baseline levels of s-CC and increasing abdominal obesity shown in our original manuscript (abdominal obesity was the only of the five components of MetS significantly associated with baseline levels of s-CC) , we also reperformed the linear regression analysis regarding s-CC influence on the progression of abdominal obesity over time entering insulin resistance as a covariate on top of all the others (except for glucose because it is included in the HOMA-IR model). Interestingly, in this model, s-CC was still highly significantly associated with increased waist circumference progression (β = 0.039, P = 0.002), although insulin resistance itself was significantly associated with waist circumference (β = 0.071, P = 0.030). In our opinion, these results highlight an independent influence of a cystatin C on abdominal obesity, thereby signifying a potential mechanism for the cystatin C-derived risk of future MetS and CVD development.
Conflict of interest statement
No conflict of interest was declared.