Freezing method affects the concentration and variability of urine proteins and the interpretation of data on microalbuminuria

Abstract

Aims Microalbuminuria and, to a lesser extent, renal tubular proteins are widely used in the early detection of incipient nephropathy in diabetes mellitus. Recent reports have indicated detrimental effects of storage at −20°C on urine proteins. This study investigated the effects of storage on the measurement of urine proteins and discusses implications for the interpretation of data. Methods Two‐hundred and sixty‐eight specimens, collected from children with Type 1 diabetes, split into duplicate aliquots and stored at −20°C and −70°C, respectively, for 6–8 months, were analysed for albumin, retinol binding protein, N‐acetyl glucosaminidase and creatinine, in the same assays to eliminate inter‐assay variability. Two independent non‐diabetic cohorts of children provided urine specimens, which were stored at −20°C for one cohort and −70°C for the other, to determine normal ranges for urine proteins. Results Storage at −20°C led to a variable underestimation of all three urine proteins in 20% of specimens. Creatinine was unaffected. This underestimation was greater in more concentrated urine (r² = 0.38, P < 0.001, n = 262). Consequently storage at −20°C increased the variance of the albumin/creatinine ratio more than the variance of albumin concentration. Temperature of storage affected the normal range, which was 0.1–2.1 mg/mmol at −20°C compared to 0.3–3.1 mg/mmol at −70°C. The prevalence of microalbuminuria (> 2 sd above the geometric mean in non‐diabetic specimens stored at −20°C) was 27% after storage at −70°C vs. 24% after − 20°C. The prevalence of microalbuminuria (> 2 sd above the geometric mean in non‐diabetic specimens stored at −70°C) was 21% after storage at −70°C vs. 17% after −20°C. Conclusions Urine proteins are significantly but variably underestimated after storage at − 20 °C. These effects account for increased variance and differences in the normal range, but have less effect on the detection of microalbuminuria than might be predicted.