Background: Quantification of proteinuria is important in the assessment of chronic kidney disease (CKD). The aim of this study was to investigate the optimal test to identify significant proteinuria.
Methods: We retrospectively assessed the relationship between total protein:creatinine ratio (TPCR), albumin:creatinine ratio (ACR) and 24-h urine total protein in 6842 patients with CKD focusing on performance at thresholds of 0.5 and 1 g/day of proteinuria.
Results: The relationship between ACR and TPCR is non-linear. TPCR is highly correlated with 24-h urine protein (Spearman's rho = 0.91), though ACR also performs well (rho = 0.84). Using receiver-operator characteristic curve analysis, TPCR outperforms ACR at predicting 0.5 g/day [area under the curve (AUC) 0.967 vs 0.951, P < 0.001] and 1 g/day of proteinuria (AUC 0.968 vs 0.947, P = 0.004). A TPCR threshold of 100 mg/mmol had a higher sensitivity (94% vs 79%) but lower specificity (88% vs 95%) than an ACR of 70 mg/mmol to predict 1 g/day of total proteinuria. To achieve comparable sensitivity, the ACR threshold falls to 17.5 mg/mmol, with lower specificity than TPCR (69.8%). Sensitivity of TPCR rose with increasing age, and in females: to achieve 95% sensitivity in a man <49 years, requires a TPCR of 65 mg/mmol, compared to 179 mg/mmol in a woman >79 years. Non-albumin proteinuria was a lower proportion of total proteinuria in patients receiving angiotensin-converting enzyme inhibitors or angiotensin receptor blockade than in those who were not (P < 0.001).
Conclusions: TPCR is a more sensitive screening test than ACR to predict clinically relevant proteinuria. The diagnostic performance of both tests varies substantially with age and gender, and should be taken into consideration when interpreting results. Total proteinuria cannot be adequately predicted from ACR, and our results suggest that caution is appropriate before utilizing ACR in patients with non-diabetic CKD.