Cystatin C-based equations for estimating glomerular filtration rate do not require race or sex coefficients

Abstract

Estimation or measurement of glomerular filtration rate (GFR) is generally required for optimal treatment of patients. Plasma creatinine has been used for estimation of GFR since 1926 and plasma cystatin C since 1979. The creatinine level is strongly dependent upon muscle mass and as the average muscle mass of different populations may vary, creatinine-based GFR-estimating equations have since 1999 used more than 10 different race coefficients to improve the diagnostic performance of such equations. But ‘race’ cannot be determined by biological measurements and is thus an ill-defined biological entity and controversial as it involves self-reporting and social considerations. In contrast, cystatin C-levels are virtually independent of muscular mass and cystatin C-based GFR-estimating equations do not require race coefficients for reliable estimation of GFR. The use of cystatin C-based GFR-estimating equations, alone or in conjunction with creatinine-based GFR-estimating equations, is therefore highly recommended to eliminate the use of race coefficients in estimating GFR. Although sex is a more biology-oriented parameter than race, sex terms may in some cases be controversial, involving self-reporting and social considerations. However, sex terms are not required for adequate estimation of GFR using cystatin C-based equations.

Keywords:

• Creatinine
• cystatin C
• glomerular filtration rate
• race
• sex

Measuring and estimating GFR

Measuring GFR always involves invasive procedures, which are time-consuming and not entirely free of risks for the patients. Some procedures use radioactive substances and should preferably not be used in children or potentially pregnant females [1]. These drawbacks of measuring GFR have prompted the development of different ways to estimate GFR. The presently most common ways to estimate GFR are based upon the plasma/serum levels of creatinine or cystatin C, or GFR-estimating equations based upon either, or both, of these levels. The diagnostic performance of creatinine- or cystatin C-based GFR-estimating equations is tested by comparing estimated GFR with GFR determined by invasive GFR-measuring procedures. It is important to realise that different invasive techniques produce slightly different results for GFR [1–3], which will influence the test of the diagnostic performance of GFR-estimating equations along with other parameters of the test procedure [4]. The general lack of this knowledge has contributed to the confusion in evaluating the merits of the multitude of GFR-estimating equations presently used [4].

Creatinine-based estimation of GFR

The introduction of endogenous creatinine clearance as a measure of GFR by Rehberg [5] initiated in 1959 the use of estimation of GFR by measuring creatinine levels in plasma or serum [6,7] and, later, by use of more complex creatinine-based GFR estimating equations [8]. Creatinine clearance has been used as a measure of GFR since 1926 [5], but it is known since 1935 [9] that it significantly overestimates GFR, since a big part of the urinary excretion of creatinine is due to tubular secretion. The insufficiency of creatinine clearance as a measure of GFR has been verified in several meta-analyses [1,10]. Already the first reports in 1959 and 1971 studying serum creatinine as a potential marker used for estimation of GFR noted that increasing muscle mass produced increasing serum levels of creatinine and that a sex coefficient had to be used to estimate GFR in males and females [6,7]. The sex coefficient used was based upon the assumption that the muscle mass, and hence creatinine production, in females was only 80–90% of that in males of the same weight and age [6,7]. Direct measurements of muscle mass and creatinine also show a strong positive correlation [11,12]. A large number of studies have shown that muscle mass strongly influences the level of creatinine, but not that of cystatin C [12–18]. Some studies show that for healthy people with normal GFR, the creatinine level is significantly correlated to the muscle mass of the individuals [11,12], but not to their measured GFR [12,18]. Agreeing with these observations, several studies indicate that the creatinine/cystatin C-ratio is an efficient marker of muscle mass [19–21].

Introduction of race coefficients in creatinine-based GFR-estimating equations

In 1999, the first creatinine-based GFR-estimating equation using a race coefficient, the MDRD equation, was introduced based upon studies of a heterogeneous American population [22]. The coefficient 1.180 was used for ‘black’ people in the equation when a coefficient of 1 was used for ‘non-black’ people. The coefficient for ‘black’ people was changed to 1.212 in the subsequent MDRD 4-variable GFR-estimating equation using creatinine values standardised to reference creatinine methods [23]. The CKD-EPI_creatinine GFR-estimating equation, described in 2009, used the coefficient 1.159 for ‘black’ people when a coefficient of 1 was used for ‘white or other’ people [24]. Differences in creatinine levels between races have been ascribed to racial differences in average muscle mass [22,25–28]. Although the MDRD equation was the first GFR-estimating equation to use race coefficients, at least 10 different race coefficients have subsequently been suggested for use in creatinine-based GFR-estimating equations (Table 1). It should also be observed that the race coefficient for ‘black’ people was generated by studies of African Americans, and thus might not be applicable for other populations with dark skin colour, e.g. in Ethiopia or Somalia. It has been clearly demonstrated that African Europeans require another coefficient for optimal results in creatinine-based GFR-estimating equations than African Americans [37], probably due to that they represent other African populations than those related to the African Americans. It is also important to realise that skin colour is not generally associated with the muscle mass of an individual. For example, different Indian Caucasian populations with dark-coloured skin are not known to be associated with a particularly high muscle mass of the individuals.

Table 1. Race coefficients used in creatinine-based GFR-estimating equations.

Population	Race coefficient or adjusted equation	Reference number
Black American	1.180	[22]
Black American	1.212	[23]
Black American	1.159	[24]
Chinese	1.233 or 1.211	[29]
Arab	1.0446	[30]
Japanese	0.808	[31]
Korean	0.73989, 0.99096, 0.9554, 1.09825, 1.04334	[32]
Thai	1.129	[33]
Turkish	0.804	[34]
Pakistani	$0.686\mathrm{\times }{\text{eGFR}}_{{\text{CKD}-\text{EPI}}^{1.059}}$	[35]
Chinese	$\text{eGFR}=627.2781\mathrm{\times }{\text{SCr}}^{-0.38089}\mathrm{\times }{\mathrm{Age}}^{-\mathrm{0.18724}}\mathrm{\times }{\mathrm{0.9286438}}^{\mathrm{if female}}$	[36]

Cystatin C-based estimation of GFR

Cystatin C was identified as a marker of GFR in 1979 [38–40]. In a study from 1985 [39], measured GFR (Cr-EDTA-clearance) and the reciprocals of the serum levels of cystatin C or creatinine were closely correlated. For creatinine, the correlations between males and females differed significantly, whereas this was not the case for cystatin C [39]. The muscle mass difference between males and females, therefore, did not seem to significantly influence the cystatin C level in contrast to the creatinine level. As mentioned above, several investigations have shown that muscle mass only marginally influences the level of cystatin C in contrast to that of creatinine [12–18]. Differences in average muscle mass between races should therefore not markedly influence cystatin C-based GFR-estimating equations. When a cystatin C-based GFR-estimating equation, CAPA, was developed in a large mixed population of Caucasians, Asians, children, and adults [41] using the international calibrator of cystatin C [42], no significant gain in diagnostic performance was achieved by the introduction of race coefficients for Caucasians or Asians [41]. When the CAPA-equation was tested in an African American population, simultaneously with another cystatin C-based GFR-estimating equation without race coefficients, CKD-EPI_{cystatin C}, the diagnostic performance for both equations was also excellent [43], strongly suggesting that cystatin C-based GFR-estimating equations will be generally useful in most, or all, populations without using race coefficients. The introduction of a sex coefficient in the CAPA-equation did not significantly improve its accuracy [41] and such a coefficient was therefore not included in the equation. Two recent studies corroborate the previous observations that cystatin C-based equations without race coefficients show smaller differences between race groups than creatinine-based equations without race coefficients [44,45].

The production of cystatin C measured as mg/min/1.73 m² has been described to be the same in men and women [46] and the plasma or serum levels of cystatin C only differ marginally between males and females [12,17,18,41,43,47,52,53,56–58]. Several cystatin C-based GFR-estimating equations without sex or race coefficients have therefore been suggested and used in clinical practise (Table 2).

Table 2. Cystatin C-based GFR-estimating equations without race or sex coefficients.

Equation	Population	References
$$\frac{162}{\text{cystC}}-30$$	Caucasian (Children)	[47]
$$\text{log}\hspace{0.17em}x\text{GFR}=1.962+\left[\mathrm{1.123 x}\hspace{0.17em}\text{log}\hspace{0.17em}x\frac{1}{\mathit{cystC}}\right]$$	Caucasian (Children)	[48]
$$-4.32+80.35\mathrm{\times }\frac{1}{\text{cystC}}$$	Caucasian (Adults)	[49]
$${99.43\mathrm{\times }\text{cystC}}^{-1.5837}$$	Caucasian (Children, Adults)	[50]
$$\frac{\mathrm{124}}{\mathrm{cystC}}-22.3$$	Caucasian (Adults)	[46]
$$84.69\mathrm{\times }{\mathrm{cystC}}^{-\mathrm{1.680}}\mathrm{\times }{\mathrm{1.384}}^{\mathrm{if}\mathit{\text{<14}}\mathrm{years}}$$	Caucasian (Children, Adults)	[51]
$$89.126\mathrm{\times }{\mathrm{cystC}}^{\mathrm{-1.675}}$$	Caucasian (Adults)	[52]
$$\text{Native}\mathrm{ }\text{CKD:}\mathrm{ }66.8\mathrm{\times }{\text{cystC}}^{-1.30}$$ $$\text{Transplant}\mathrm{ }\text{recipient:}\mathrm{ }76.6\mathrm{\times }{\text{cystC}}^{-1.16}$$	US adults	[53]
$$79.901\mathrm{\times }{\text{cystC}}^{-1.4389}$$	Caucasian	[54]
$$130\mathrm{\times }{\text{cystC}}^{-1.069}\mathrm{\times }{\text{age}}^{-0.117}-7$$	Caucasian, Asian, African American (Children, adults)	[41]
$$91\mathrm{\times }{\text{cystC}}^{-1.213}$$	Caucasian (Children)	[55]
$$\frac{\mathrm{107.3}}{\left(\frac{\mathit{cystC}}{\mathit{QcystC}}\right)}\mathrm{\times }{\mathrm{0.988}}^{\mathit{\text{Age-40 (if>40years)}}}$$ QcystC = 0.82 if age <70 years or QcystC = 0.95 if age >70 years	Mainly Caucasian (Children, Adults)	[58]

The equations of references [41,55,58] are based upon the international cystatin C calibrator [42].

Race and sex coefficients should, and can, be avoided in GFR-estimating equations

The first race coefficient in a GFR-estimating equation was introduced in 1999 when a coefficient for ‘black’ people in USA was introduced in the creatinine-based MDRD equation [22]. Since then, at least 10 different race coefficients have been introduced in creatinine-based GFR-estimating equations for different populations in at least nine nations [22–24,29–36]. One of these equations is the frequently used CKD-EPI_creatinine GFR-estimating equation [24]. However, ‘race’ cannot be determined by biological measurements [59–62] and is usually based upon self-reporting [63] and the concept of race is therefore subjective and more associated with sociological circumstances than objective biological facts [63]. It is also associated with discomfort of the self-reporting individual and misclassification may lead to reduced access to care [63]. Therefore, strong recommendations have recently been issued that race coefficients should not be used in creatinine-based GFR-estimating equations [44,45,64] and that cystatin C-based equations, which do not require race coefficients, can replace the creatinine-based equations [44,45,64]. New creatinine-based GFR-estimating equations without the use of race coefficients have been created, but their diagnostic performance seems to be inferior to that of cystatin C-based equations in populations with non-homogeneous ethnicity [44,45,64].

The use of race coefficients in creatinine-based equations is obviously problematic and should be avoided, but the use of sex coefficients might also be associated with problems. Although it for many years has been assumed that the sex of a person always is a binary parameter, recent research has demonstrated that this is not the case [65–67]. The construction of creatinine-based GFR-estimating equations has since its beginning always assumed that sex is a binary parameter and thus used different coefficients for the female sex in all equations, in which male sex routinely has been assigned the coefficient 1. This way of constructing GFR-estimating equations based upon creatinine has later been implemented in the construction of cystatin C-based equations, although it is known that sex differences between men and women concerning the cystatin C level is small [12–18,41]. It is therefore possible to create cystatin C-based GFR-estimating equations with good diagnostic performance without using sex coefficients [41,46–55,58]. When some of these equations were constructed, the difference between measured and estimated GFR using equations with and without sex coefficients was investigated and found to be about 1% or not statistically different from 0 [41,46,53]. Another problem with the use of a binary sex coefficient is that several nations, e.g. Germany, Canada, India, Australia, Denmark, have applied the results of recent research and specified more than two genders in basic descriptions of their citizens [68]. Self-reporting of sex is often used in health care and if only two alternatives are allowed, a significant proportion of the patients will experience discomfort connected to the acknowledgment of more than two genders in national legislations and issues pertaining to the LGBTQIA + spectrum [69,70]. All the above-mentioned problems associated with the use of binary sex coefficients can be avoided if cystatin C-based GFR-estimating equations are used. This strongly supports the notion that estimating GFR using cystatin C should be a basic resource in clinical routine and not only used when the results of creatinine-based GFR-estimating equations are questionable.

It should be observed, that although cystatin C, in contrast to creatinine, is virtually independent of muscle mass allowing construction of cystatin C-based GFR-estimating equations without race and sex coefficients, creatinine is in some cases less influenced by non-renal factors than cystatin C [71]. For example, glucocorticoid treatment raises the level of cystatin C by increasing the production of it, whereas it does not affect the creatinine level [71]. The cost for analysis of cystatin C is also generally higher than that for analysis of creatinine. But the difference in cost between automated immunochemical analysis of cystatin C and automated enzymatic analysis of creatinine is not very big.

Disclosure statement

The authors report no conflict of interest and are alone responsible for the content and writing of the article.

Correction Statement

This article has been republished with minor changes. These changes do not impact the academic content of the article.

Additional information

Funding

The investigation was supported by grants from Alfred Österlund Foundation, Skåne University Hospital Funds, and the Medical Faculty of Lund University.