Article Text
Abstract
Background The 24 hr urine (24H-P) collection sample is the gold standard test for proteinuria measurement in lupus but this test is cumbersome for the patients. Spot Urine Protein Creatinine Ratio (S-UPCR) in single voided urine sample has been accepted to measure proteinuria instead of 24H-P. There is no agreement amongst the existing studies on the utility of S-UPCR vs. 24H-P in screening/monitoring proteinuria. The variance between the studies is related to the use of inappropriate statistical analyses.
Objectives To determine the utility of S-UPCR as a screening and diagnostic test in the assessment of proteinuria in lupus.
Methods This is a retrospective analysis performed on the data from a single lupus cohort on patients seen between May 2008-December 2014. Inclusion criteria: Laboratory records with non-missing data for S-UPCR and 24H-P done at the same time. Exclusion criteria: Patients with diabetes mellitus, end stage kidney disease and kidney transplantation. The main variables were 24H-P, S-UPCR and the ratio of 24H-P and creatinine (24H P/C).
Laboratory records were divided into 4 groups according to 24H-P: Group 1: <0.5, group 2: 0.5-0.99, group 3:1-1.99, and group 4:≥2 g/day. Descriptive statistics were used for continuous/ categorical variables. Pearson correlation coefficient was measured for 24H-P and S-UPCR. Agreement was determined by Inter Class Correlation Coefficient (ICC), Concordance Correlation Coefficient (CCC) and Bland-Altman plot between 24H P/C and S-UPCR.
The best cut off for S-UPCR comparable for 24H-P of 0.5 g/day was determined with ROC curve and linear regression (controlling for age and sex).
Results 1730 laboratory records from 421 patients were retrieved; 85.3% women, 52.5% Caucasian and mean disease duration 11.61±9.56 years.
24H-P and U-SPCR correlation: For all records n=1730 r =0.73; group1 n=923 r=0.3; group 2 n=297 r=0.2; group 3 n=246 r=0.2; and group 4 n=264 r=0.5 (all p significant). Although the correlation of all records is strong, the results of the correlations among different groups showed weak to moderate correlation.
ICC(2, k) for 24H P/C and U-SPCR: All records ICC=0.85; group 1 ICC=0.41; group 2 ICC=0.67; group 3 ICC=0.62 and group 4 ICC=0.78. CCC for 24H P/C and U-SPCR: All records CCC=0.84; group 1 CCC=0.47; group 2 CCC=0.57; group 3 CCC=0.55 and group 4 CCC=0.81. ICC and CCC should have exceeded 0.9 to ensure reasonable validity.
Bland-Altman plot: For proteinuria ≥0.5-2g/day (groups 2, 3 and 4), the agreement between 24H P/C and U-SPCR is poor (Figure 1). U-SPCR overestimated the amount of proteinuria compared to 24H-P.
ROC curve: U-SPCR of 0.08 g/mmol (800 mg/mmol) has the best sensitivity (93%) and specificity (72%) for screening for proteinuria when compared to 24H-P (cut of 0.5 g/day). The linear regression analysis also identified 0.08 as the best cut off.
Conclusions The correlation/agreement (Pearson, ICC, CCC and Bland-Altman) analyses confirmed that U-SPCR has no reasonable validity in accurately measuring proteinuria compared to 24H-P. The results of the ROC showed that U-SPCR can be used as a screening test (qualitative test) and the best cut off for 24H-P of 0.5 g/day is 0.08. 24H-P should be the gold standard test to accurately measure proteinuria especially to follow a patient's response to therapy.
Disclosure of Interest None declared