Predictors of serum creatinine in haemodialysis patients: a cross-sectional analysis.

BACKGROUND

C-reactive protein (CRP) levels, an acute phase response index, predict cardiovascular outcome and are inversely related to visceral proteins, including albuminaemia in haemodialysis patients. Less definite is the relationship between inflammation and markers of somatic proteins such as serum creatinine in such patients. To explore these questions, a cross-sectional analysis of potential predictors of serum creatinine was performed.

METHODS

One hundred and seventy-nine prevalent haemodialysis patients as of June 2001 were included in the cohort. Midweek pre-dialysis blood samples were collected during the months of June, September through to December 2001 inclusive, and determinations of serum urea (urease method), creatinine (alkaline picrate method) and CRP levels by means of a high sensitivity immunonephelometric method were performed. Furthermore, pre- and post-dialysis body weights were recorded and 2 min post-dialysis serum urea levels were determined three times. They were utilized for the calculation of single pool Kt/V and of normalized protein catabolic rate (nPCR). Each of the data represents the mean of three determinations made every 3 months in the study period.

RESULTS

The analysis of multivariate linear regression was able to validate our model characterized by a dependent variable, serum creatinine and four independent variables (age, CRP, Kt/V and nPCR) (R(2)=0.60; F=24.10; P<0.00001; SE=1.94). Age (-0.08 mg/dl decrease in serum creatinine per 1-year increase in age), Kt/V (-0.25 mg/dl decrease in serum creatinine per 0.1 increase in Kt/V) and nPCR (0.10 mg/dl increase in serum creatinine per 0.1 g protein/kg/day increase in nPCR) were independently predictive of serum creatinine (P<0.00001). CRP and dialysis vintage did not predict serum creatinine. Stratifying the patients for the effects of CRP, only CRP values </=4 mg/l were directly predictive of serum creatinine (P<0.00001), whereas CRP values >4 mg/l were not. A further insight was given by the stratification of the patients for the effects of the interquartile ranges of CRP: it showed a progressive and statistically significant reduction of beta-coefficient inversely related to the increasing CRP values (P=0.003). Thus, the nature of the correlation between CRP and serum creatinine changes with increasing CRP values: from being a direct one, it shows a trend towards a transformation into an indirect one with beta=0 at a CRP value of approximately 9 mg/l. However, this indirect relationship does not reach statistical significance.

CONCLUSIONS

The present cross-sectional study suggests that the activation of acute phase response does not influence creatinine metabolism in haemodialysis patients; in contrast, age, Kt/V and nPCR predict serum creatinine levels. Larger prospective trials are needed to achieve a definitive answer about the relationship between somatic proteins, acute phase response activation and nutrition in dialysis patients.