In the present study, we observed an association between increased serum IGF-I and decreased eGFR in men but not in women and between increased serum IGFBP-3 concentrations and decreased eGFR in both sexes. Our findings regarding IGF-I confirm the data from the NHANES III study  revealing a positive association between serum IGF-I concentrations and risk of CKD.
A study in monozygotic and dizygotic adults twin pairs of the same sex showed that women had higher serum IGFBP-3 concentrations than men . Furthermore, sex-specific differences in serum IGF-I concentrations were detected and showed that the median serum IGF-I value was 3 ng/mL lower in women compared to men [14, 15]. The differences in serum IGF-I or IGFBP-3 concentrations between men and women justify performing sex-segregated analysis tests and may explain the slightly different results with respect to the association between IGF-I and CKD in the present study. However, the positive association between increased serum IGF-I and decreased eGFR only barely missed statistical significance in women. The same trend was observed in NHANES III .
In comparison to our results, a small Danish study found lower free serum IGF-I and higher serum IGFBP-3 concentrations in patients with CKD compared to healthy controls . However, total IGF-I concentrations were slightly but not significantly increased in patients with CKD compared to healthy controls. Peritoneal dialysis and haemodialysis patients had significantly higher serum IGF-I and IGFBP-3 concentrations than subjects with normal renal function . Another study  showed normal serum IGF-I concentrations, but elevated serum IGFBP-3 concentrations in patients with CKD and different types of renal osteodystrophy compared to healthy control subjects. Taking into account the decreased renal clearance and thus an increased half-life of IGF-I in CKD , elevated serum IGF-I concentrations may be expected and normal serum IGF-I concentrations may indicate reduced production. In advanced renal failure, other condition such as metabolic acidosis has been shown to reduce serum IGF-I concentrations . The altered metabolic milieu in severe CKD affects the secretion of hormones and the hormone-induced response of target tissues, causing endocrine dysfunctions .
A further study among 500 never-treated hypertensive patients  showed an positively and strong association between IGF-I and eGFR which is in contrast to our results even if we have a high proportion of subjects with hypertension. Possible reason for this discrepancy might be that compared to the never-treated hypertensive subjects investigated by Peticone et al.  the large part of our participants with hypertension (44.9%) were treated. There is strong evidence for an association of renal function and cardiovascular mortality even in non-dialysis-dependent CKD , which was confirmed by a recent large meta-analysis of 21 general population cohorts . This association might be modulated by different IGF-I serum concentrations. In previous epidemiological studies IGF-I, as well as IGFBP-3, has been suggested as risk factor for all-cause and cardiovascular mortality . In a recent meta-analysis  low, as well as high, IGF-I serum concentrations were associated with increased mortality in the general population. Thus, in early or moderate renal failure, serum IGF-I concentrations may be elevated due to increased IGF-I half-life and may become clinically relevant in later stages of renal failure. Nevertheless, we assume that the IGF/GH axis is a biomarker rather than a risk factor for renal diseases and reflects the healthy status which might also be influenced by renal disorders. Therefore, IGF-I is not specific for impaired renal function.
IGFBP-3 is specifically cleaved by proteases and low-molecular weight IGFBP-3 fragments accumulating in CKD may be detected by the usual radioimmunoassay. In an earlier study in CKD patients, the intact serum IGFBP-3 concentrations stayed within normal limits .
The key strength of our study is the large population-based sample of 4028 men and women aged 20–81 years. A further strength is the high grade of quality assurance of the laboratory method. All assays were performed on a single analyzer using the same reagents and were conducted by skilled technicians and, therefore, minimizing analytical variability. However, a limitation of our study arise from the cross-sectional design that is generally not suitable to prove causal relations. Furthermore, no measurements of free IGF-I or IGF-II were available which would provide a more complete view on the relation between the IGF/GH axis and renal function.