Mind the Gap: Discordance Between Cystatin C and Creatinine eGFR and Diabetes Complications Free

Estimation of glomerular filtration rate (eGFR) from serum creatinine (eGFRcr) or cystatin C (eGFRcys), or the combination of both markers, is an established approach to measure kidney function. However, the degree of discordance between eGFRcr and eGFRcys (termed eGFRdiff) is emerging as a useful marker of health status and the risk for developing adverse events in people with diabetes, as highlighted by Harjutsalo et al. (1) in this issue of Diabetes Care.

More than 40 years ago, Grubb et al. (2) proposed that serum cystatin C could be used as a marker of glomerular filtration rate (GFR), since it appears to be influenced to a lesser extent by non-GFR factors compared with serum creatinine. Cystatin C is a low-molecular-weight (13.3-kDa) protein that is produced at a constant rate by all nucleated cells, freely filtered by the glomerulus, and then metabolized by the proximal tubule. However, like creatinine, evidence is accumulating that serum cystatin C levels are not free from the effects of non-GFR factors. Well-recognized non-GFR influences on serum creatinine are age, sex, protein intake, physical activity, and factors that affect tubular creatinine secretion. In contrast, adiposity, systemic inflammation, thyroid disease, and steroids appear to influence serum cystatin C levels (3,4). eGFRcr lacks accuracy in some clinical situations, and its limitations in the setting of diabetes are well documented (5). Generally, eGFRcys performs better than eGFRcr when compared with measured GFR, especially at high-normal GFR values, and recent guidelines have advocated for increasing the use of serum cystatin in GFR estimation equations, either alone or in those that combine both serum creatinine and cystatin C (6–8).

Given the findings described above, it is not surprising that intraindividual discrepancies between eGFRcr and eGFRcys (i.e., eGFRdiff) are reported to be common, with around one in four individuals having a difference of at least 15 mL/min/1.73 m². In the general population seeking routine medical care, large absolute differences (eGFRabdiff) or relative differences (i.e., ratio of eGFRcys to eGFRcr, or eGFRrediff) between eGFRcys and eGFRcr are defined by certain clinical characteristics, with the presence of this difference also being associated with adverse clinical outcomes (9). What is the significance of this difference in people with diabetes? Two recent publications in Diabetes Care have highlighted the clinical characteristics associated with the finding of eGRFdiff in people with type 1 or type 2 diabetes and have linked eGFRdiff with a subsequent increased risk of developing diabetes-related microvascular complications, mortality, and cardiovascular (CV) events (1,10).

Harjutsalo et al. (1) calculated GFRcys and eGFRcr, derived from the 2009 Chronic Kidney Disease Epidemiology Collaboration equations, in 3,769 individuals with type 1 diabetes participating in the Finnish Diabetic Nephropathy (FinnDiane) Study. The degree of concordance (−15 mL/min/1.73 m² < eGFRabdiff > 15 mL/min/1.73 m²) and discordance (calculated as eGFRcys – eGFRcr), defined as negatively discordant, eGFRabdiff <−15 mL/min/1.73 m² (i.e., eGFRcys < eGFRcr), or positively discordant, eGFRabdiff ≥15 mL/min/1.73 m² (i.e., eGFRcys > eGFRcr), between eGFRcys and eGFRcr was assessed. The overall concordance rate was 63%, the negative discordance rate was 20%, and the positive discordance rate was 17%.

Compared with the concordant group, those in the negative-discordance group were younger, had a lower age at onset of diabetes, had a shorter duration of diabetes, were more likely to be men, were more likely to be current smokers, had a worse lipid profile, had lower BMI, and had higher hs-CRP levels and insulin doses. Those in the positive-discordance group were more likely to be women, had higher age at onset of diabetes, had normal albuminuria, had less retinal laser treatment, were less likely to be current smokers, and had better lipid profiles. Interestingly, positive discordance rates were found to be greatest in individuals who did not have albuminuria. In this setting, eGFRcys most likely better reflects true GFR and its changes in the high-normal range, kidney function that would be expected for people with type 1 diabetes and normoalbuminuria, compared with measures of kidney function derived from creatinine (11,12). In addition, this study showed that both increases in baseline serum cystatin C and decreases in baseline eGFRcys were independently associated with the development of moderate albuminuria (microalbuminuria) over 11 years of follow-up. In comparison, baseline eGFRcr was not associated with the development of moderate albuminuria.

These findings build on those from another study published in Diabetes Care in 2024 by He et al. (10). In that study, the authors assessed the significance of eGFRdiff for predicting the development of diabetes microvascular complications over a follow-up period of 13.6 years. The study examined 25,825 participants free from microvascular complications at baseline who were enrolled in UK Biobank, with the vast majority of participants presumably having type 2 diabetes.

The incidence of microvascular complications (diabetic retinopathy, diabetic kidney disease, and diabetic neuropathy) was ascertained using electronic health records. Compared with participants with concordant eGFRabdiff, the risks of diabetic retinopathy (hazard ratio [HR] 1.13, 95% CI 1.22–1.36), diabetic kidney disease (HR 1.63, 95% CI 1.50–1.76), and diabetic neuropathy (HR 1.40, 95% CI 1.23–1.58) were significantly increased in those with negatively discordant eGFRabdiff in a fully adjusted Cox proportional analysis that also controlled for multiple factors, including age, sex, HbA_1c, BMI, eGFRcr, baseline albuminuria, and CRP. Each 10% decrease in eGFRrediff was also found to be significantly associated with an increased risk for the development of the three diabetes-related microvascular complications described above. Overall, those with positive-discordance eGFRabdiff results were found to be at reduced risk for the development of diabetes-related microvascular complications.

He et al. (13) also examined the significance of the association between eGFRdiff and the subsequent risk of mortality and CV events in adults with diabetes in a study involving three large international databases (39,121 participants in total). Individuals with a negative eGFRabdiff or a lower eGFRrediff were at a significantly increased risk for the diabetes-related complications described. Recently, a low eGFRrediff (<0.6) has been associated with increased mortality in people with type 2 diabetes and in a limited number of individuals with type 1 diabetes in a community-based Australian study (14). This study included individuals who did not have reduced kidney function or albuminuria.

What factors account for a large eGFR discordance in some individuals? Sarcopenia and inflammation are likely significant non-GFR players, among other factors, explaining differences in serum creatinine and cystatin C levels, respectively. At face value, a greater proportion of people with type 2 diabetes appear to have negatively discordant eGFRdiff values than corresponding members of the general population or individuals with type 1 diabetes. There may also be differences in the clinical characteristics associated with a negative eGFRdiff in people with type 1 versus type 2 diabetes. Specific factors in the setting of type 1 diabetes may influence muscle health, and hence serum creatinine levels, and may account for the above finding. The shrunken-pore hypothesis has also been proposed as an explanation for the discordance between eGFR derived from cystatin C and that derived from creatinine. This is a phenotype of kidney dysfunction where there is a selective decrease in the glomerular filtration of larger molecules (cystatin C is 13 kDa) versus smaller molecules (creatinine is 0.13 kDa) (15). At present, this hypothesis remains speculative, but there is some evidence linking thickening of the glomerular basement membrane and eGFRrediff (16).

In summary, a negative eGFRdiff could be a valuable prognostic tool, with a greater discrepancy (eGFRcr > eGFRcys) linked to poorer metabolic profiles and increased risk for the development of microvascular complications and CV events in people with diabetes. Integrating cystatin C measurements into clinical practice could enhance risk assessment and management strategies for people with diabetes. The significance and delineation of the factors accounting for a negative eGFRdiff would be facilitated by serial studies examining a potential widening or narrowing of eGFR discordance and its association with clinical and metabolic factors, together with the ability to predict adverse outcomes. Studies that incorporate a direct measure of GFR and those that also aim to determine the contribution of glomerular versus nonglomerular factors to the discordance in eGFRcys or eGFRcr would also be useful in this area.