Aligning Albuminuria and Proteinuria Measurements

Proteinuria is central to the diagnosis, staging, and management of CKD. CKD guidelines in adults stage disease using albuminuria categories focusing on spot urine albumin-to-creatinine ratios (ACRs).¹ However, measurement of total protein-to-creatinine ratio (PCR) is widespread in both research and clinical settings, and hence the ability to understand how average and individual PCR levels relate to ACR levels is important. Although the guidelines provide a table relating PCR to ACR categories, there has been a longstanding need for better quantification of this relationship, its nonlinearity, modifying influences, and precision.

Weaver et al.² provide a rigorous analysis of a large clinical database with 47,714 measurements of both PCR and ACR on adult patients from Alberta, Canada. Their results are consistent with previous research in the area but provide an unprecedented rigor and sophistication in quantifying the relationship. They show that albumin comprises <30% of urinary protein in CKD stage A1 (ACR<30 mg/g) compared with >60% at stage A3 (ACR>300 mg/g). As a result, the relationship between PCR and ACR is nonlinear, flatter, and less correlated at lower levels of PCR. In fact, at PCR<50 mg/g there is nearly no relationship between PCR and ACR. At higher levels there is an approximately linear relationship on the log-log scale. They provide a detailed analysis using a quantile regression with both cubic splines as well as linear splines.² The latter provide an excellent approximation that can be easily understood and implemented. They confirm that the Kidney Disease Improving Global Outcomes suggestion for approximating the borders of CKD stage A2 (ACR cut-offs of 30–300 mg/g) using PCR of 150–500 mg/g. Their regressions show a PCR of 150 mg/g corresponds to a median (interquartile range [IQR]: 25th–75th percentile) ACR of 35 (IQR, 16–66) mg/g and a PCR of 500 mg/g corresponds to a median ACR of 301 (IQR, 213–357) mg/g. The wide IQR demonstrates that although the average relationship of PCR to ACR is strong, prediction for individual patients is imprecise (95% coverage intervals would be approximately 2.4 times wider than interquartile intervals). They show that the association of PCR to ACR is modified by a number of factors, with sex being the most important (women having lower ACR at the same PCR), and CKD stage A2 being the range where differences are most noticeable. Classification of patients into ACR categories benefits from converting PCR to estimated ACR, rather than simply using PCR categories. However, sensitivity for detecting stage A2 is still limited. Finally, the paper includes follow-up data on risk of ESKD, which allows for a comparison of risk prediction using the Kidney Failure Risk Equation (KFRE) and measured ACR versus ACR estimated from PCR. In this prediction context, the two perform similarly, suggesting that for ESKD risk estimation, appropriately converted PCR data can be quite useful, perhaps because higher levels are more predictive of ESKD risk and are most correlated to ACR levels. It is worth exploring how well conversion performs in additional risk algorithms that include albuminuria (www.ckdpcrisk.org) and can allow for risk stratification, which may aid in prioritizing effective but costly treatment strategies.

The study by Weaver et al. benefits from having a large data set and a rigorous thoughtful analysis. Its main limitation is generalizability and the authors call for replication in additional settings. Requiring ACR and PCR measures on the same day limited the data to 4.4% of ACR measurements in Alberta, Canada. Most measurements were made in two large clinical laboratories. The extent to which the results would apply to other settings and laboratory methods for measuring ACR and PCR should be explored to determine the qualitative and quantitative generalizability of the results.

It is useful to consider the implications of the study for when ACR should be measured versus estimated from PCR. The article provides a useful summary of potential applications for PCR over ACR conversion,² recommending use in retrospective clinical and research settings when PCR data are available and ACR data are not. This recommendation aligns with the guideline preference of ACR over PCR measurement in adults. Assays of all protein in the urine suffer from heterogeneity leading to a lack of gold-standard or reference methods. In contrast, because albumin is a single protein, reference methods and materials have been developed as a part of an ongoing effort to improve standardization.^3,4 College of American Pathologists survey data suggests that most ACR methods are precise (coefficients of variation of 5%–8%) but biases can still be large, averaging 40% compared with a liquid chromatography–tandem mass spectrometry reference method. Thus, further work is needed, but success in standardization of serum creatinine provides a useful model. In addition to analytical variability, albuminuria suffers from high intraindividual variability (coefficient of variation of 30% for ACR),⁵ which can be ameliorated by averaging multiple measurements. Despite imperfect standardization and high variability, albuminuria is a strong risk factor for a wide range of future events,¹ and recently, change in albuminuria has been shown to be a useful surrogate outcome in clinical trials of CKD progression.⁶ Thus, albuminuria is the recommended and preferred method for staging CKD and following progression, although its cost is higher and availability in low- and middle-income countries is low.⁷

In summary, albuminuria is central to CKD diagnosis, staging, and management. Efforts to improve ACR standardization and use should continue. Proteinuria quantification will likely never be standardized, but Weaver et al. provide useful methods for aligning existing PCR data with corresponding ACR levels. If proven to be generalizable, their results can be used to harmonize existing PCR data with CKD staging and risk algorithms that focus on albuminuria.

Disclosures

Dr. Coresh receives grants from the National Institutes of Health and National Kidney Foundation, and is a scientific advisor to Healthy.io.