Figures
After publication of this article [1], it came to light that there were errors in the reported glomerular filtration rate (GFR) estimates.
The two-fold purpose of this paper [1] was to 1) compare accuracy and bias of widely used glomerular filtration rate (GFR) estimating equations to a gold-standard GFR measure (iohexol disappearance from plasma) in HIV-positive and HIV-negative volunteers, and 2) to assess factors associated with bias and accuracy of the creatinine-based and cystatin C-based equations. Recently, our co-investigators, who performed the laboratory analyses and calculations for the iohexol GFR, identified a drift that occurred in their measurement of iohexol (prior to this study) that led to an across-the-board underestimation of iohexol concentrations from blood samples, which produced a systematic overestimation of GFR by approximately 10%. This measurement error in this laboratory was described in a publication in 2017[2]. We subsequently repeated the analyses in the PLOS ONE paper using recalibrated (corrected) iohexol GFR values provided here in an updated version of Table 1.
Because mGFR was recalibrated approximately 10% lower and mGFR was central to analyses, almost all estimates in Table 2, Table 3, and Table 4 have been revised, with substantive changes described below. We also revised all Figs 1–3, although the clinical inferences from the figures are unchanged.
Bland-Altman plots for estimated and measured glomerular filtration rate (GFR) in HIV-positive participants using the CKD-EPI equations for serum creatinine (A), cystatin C (B), or both biomarkers (C). The average GFR (measured and estimated) is shown on the X axes. Bias, defined as the difference between estimated and measured GFR, is displayed on the Y axes. The average biases are represented by the horizontal solid lines and the horizontal dashed lines represent 2 standard deviations above and below the averages.
Bland-Altman plots for estimated and measured glomerular filtration rate (GFR) in HIV-negative participants using the CKD-EPI equations for serum creatinine (A), cystatin C (B), or both biomarkers (C). The average GFR (measured and estimated) is shown on the X axes. Bias, defined as the difference between estimated and measured GFR, is displayed on the Y axes. The average biases are represented by the horizontal solid lines and the horizontal dashed lines represent 2 standard deviations above and below the averages.
Correlation of estimated glomerular filtration rate (eGFR) bias, defined as the difference between eGFR and measured GFR, with percentage of activated CD8 T cells (CD38+ and HLA-DR+) using the creatine-based CKD-EPI equation in HIV-negative (A) and HIV-positive (B) subjects, and the cystatin C-based CKD-EPI equation in HIV negative (C) and HIV-positive (D) subjects. The percentage of CD8+ T cells with an activated phenotype is shown on the X axes (note, different scales for HIV-positive and HIV-negative groups). Rho is the spearman rank correlation coefficient, which may vary between -1 and 1. The dashed lines represent least-squares regression lines.
- In the original paper, we reported that the cystatin C-based equation (eGFRcys) was the least accurate and most biased of the three CKD-EPI equations in HIV-positive participants. In the revised analysis, we found that the creatinine-based equation (eGFRcr) was the least accurate and most biased of the three equations. This is relevant because eGFRcr is the most commonly used equation in clinical practice. Consistent with the original analysis, the combined biomarker equation (eGFRcr-cys) remained the most accurate and least biased equation.
- In contrast to the original analysis, we found that the accuracy and bias of eGFRcr varied significantly by stratum of mGFR (<90 vs. ≥90 mL/min/1.73m2) in both the HIV-positive and HIV-negative groups, such that this equation was more biased and less accurate at lower levels of kidney function than at higher kidney function. This is important, because accurate GFR estimation may be more important at lower compared with higher levels of kidney function.
- Consistent with the original analysis, we found that the bias of eGFRcys was influenced by immune activation and HIV viremia, whereas eGFRcr performance was not affected by these factors (Fig 3). However, in contrast to the original analysis, these factors were no longer statistically significantly associated with the accuracy of eGFRcys.
Please see the revised Figs 1–3 and revised Tables 2–4 here.
A member of PLOS ONE's Editorial Board reviewed the new results and underlying data and confirmed that they support the overall conclusions reported in the article.
Supporting information
S1 File. Study Dataset.
Clinical Variables eGFR and mGRF.
https://doi.org/10.1371/journal.pone.0215630.s001
(DTA)
References
- 1. Bhasin B, Lau B, Atta MG, Fine DM, Estrella MM, Schwartz GJ, et al. (2013) HIV Viremia and T-Cell Activation Differentially Affect the Performance of Glomerular Filtration Rate Equations Based on Creatinine and Cystatin C. PLoS ONE 8(12): e82028. pmid:24376511
- 2. Schwartz GJ, Wang H, Erway B, et al. Multicenter Laboratory Comparison of Iohexol Measurement. The Journal of Applied Laboratory Medicine: An AACC Publication 2017: jalm. 2017.024240.
Citation: Bhasin B, Lau B, Atta MG, Fine DM, Estrella MM, Schwartz GJ, et al. (2019) Correction: HIV Viremia and T-cell Activation Differentially Affect the Performance of Glomerular Filtration Rate Equations Based on Creatinine and Cystatin C. PLoS ONE 14(4): e0215630. https://doi.org/10.1371/journal.pone.0215630
Published: April 16, 2019
Copyright: © 2019 Bhasin et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.