Kidney Week

Abstract: TH-PO998

Estimating GFR from Cystatin C Without Including a Sex Variable: CKD-EPI 2023 Equation

Session Information

• CKD Epidemiology, Risk Factors, Prevention - I
  November 02, 2023 | Location: Exhibit Hall, Pennsylvania Convention Center
  Abstract Time: 10:00 AM - 12:00 PM

Category: CKD (Non-Dialysis)

• 2301 CKD (Non-Dialysis): Epidemiology, Risk Factors, and Prevention

Author

• Inker, Lesley Ann, Tufts Medical Center, Boston, Massachusetts, United States

Lesley Ann Inker,

Group or Team Name

• CKD-EPI.

Background

Use of a binary sex variable in eGFR equations may limit their use in gender subgroups. There are smaller differences between sex groups in cystatin C than creatinine independent of GFR. Other research groups have developed equations to estimate GFR using cystatin without sex [Pottel et al, NEJM 2023; Grubb et al, Clin Chem 2014]. We developed a CKD-EPI cystatin C equation that does not include a variable for sex and evaluated its performance in an external validation population.

Methods

Using the same population (5,352 participants from 13 studies) used for development of the current equation that includes age and sex (CKD-EPI 2012 eGFRcys AS), we developed a new eGFR cystatin C equation without sex (CKD-EPI 2023 eGFRcys A). We used least squares linear regression of measured GFR (mGFR) vs serum cystatin C and age on the logarithmic scale, with separate slopes for cystatin C at low vs. high values using the same spline knot as the 2012 equation (0.8 mg/L), and assessed model performance using root-mean-square error (RMSE). We assessed performance in the CKD-EPI 2021 external validation population (4,050 participants from 12 studies) using bias (systematic error, mL/min/1.73m²) as the median difference between mGFR and eGFR, and accuracy as the percentage of eGFR within 30% of mGFR (P₃₀).

Results

In the development and validation population, 2,245 (42%) and 1,557 (38%) were female, respectively. Removing the sex variable led to a minimal increase in RMSE overall and in both sex subgroups (Table). In the external validation population, removing the sex variable led to a small increase in bias in both subgroups, with greater decrease in accuracy for females vs males although at levels considered acceptable (P₃₀ >80%).

Conclusion

The availability of acceptably accurate eGFRcys equations that do not include a sex variable provides an option to use in people whose gender identification differs from their sex assigned at birth. Further studies can explore the impact of using these equations in the general population.

Funding

• NIDDK Support