Introduction
Glomerular filtration rate (GFR) is a validated marker of kidney function [
1,
2] recommended in treatment guidelines [
3,
4]. In patients with neuromuscular diseases muscle mass is diminished and therefore creatinine-based estimations systematically overestimate kidney function and reduced kidney function may remain undetected and undiagnosed [
5,
6]. When serum creatinine is combined with information on lean body mass, a slightly better estimate is achieved [
7]. Cystatin C (CysC) correlates well with the clearance of inulin,
51Cr-EDTA and iohexol, and is not affected by muscle mass to the same extent as creatinine [
8,
9] and has therefore been suggested as a better marker of kidney function [
10‐
12]. CysC has been evaluated in different patient populations, but few studies have assessed CysC in patients with primary neuromuscular diseases including Duchenne muscular dystrophy (DMD) [
13,
14], amyotrophic lateral sclerosis (ALS) [
15] and myotonic dystrophy type 1 (DM1) [
16]. Recent studies have shown that CysC concentrations are affected by determinants other than kidney function and muscle mass. Age, male gender, body mass index (BMI), fat mass, triglycerides, hypertension, uric acid, C-reactive protein and diabetes have all been associated with higher serum concentrations of CysC regardless of kidney function [
12,
17]. This is of importance since patients with muscular dystrophy often have an absolute or relative increase in body fat [
18,
19]. Treatment with high doses of glucocorticoids, as well as thyroid dysfunction and ethnicity also influence CysC levels [
10‐
12]. A report from the Swedish Council on Health Technology Assessment highlighted that cystatin C-based equations had not been sufficiently evaluated in patients with a low BMI [
20].
This study was designed to evaluate the precision, accuracy, and bias of two creatinine-, one cystatin C-based and one combined equation to estimate glomerular filtration rate (eGFR) in patients with primary neuromuscular diseases. The findings may also be valuable for subjects with low muscle mass or BMI due to other reasons [
21].
Discussion
In this study we show that s using a cystatin C-based equation to estimate kidney function in patients with primary neuromuscular disease results in better precision, accuracy and lower bias compared to creatinine-based estimations, but most importantly, we show that all creatinine- and cystatin C-based equations evaluated in this study overestimate kidney function in this patient population and especially in patients with reduced kidney function. In addition, and in contrast to other patient populations, a combined equation based on the mean values of creatinine and cystatin C is not more precise and accurate than a cystatin C-based estimation [
30].
Previous studies have demonstrated the potential usefulness of cystatin C as a marker of kidney function in patients with Duchenne muscular dystrophy [
13,
14], amyotrophic lateral sclerosis [
15] and myotonic dystrophy 1 [
16], but in a study in children and adolescents with spinal dysraphism the authors concluded that slightly to moderately reduced kidney function could still remain undiagnosed [
31]. Interestingly, in a recent study in community-dwelling older adults with sarcopenia and chronic kidney disease eGFR creatinine was not significantly correlated to muscle volume and strength, on the contrary, eGFR cystatin C was positively correlated to these parameters [
32]. The findings in our study support guidelines recommending cystatin C-based estimation of kidney function when creatinine-based equations could be inaccurate. However, a weakness with the current guidelines is that they do not specify in which populations this should be done [
3,
4]. Even though a cystatin C-based equation may provide better accuracy, in our study it varies within a wide range especially in patients with reduced kidney function (35–67%). In addition, a bias of 22 and a low P30 of 49% could probably not be considered “good enough” for the use of these estimates in clinical practise, especially in patients where a reduction in kidney function is suspected. We therefore argue that kidney function should be measured and not just estimated in patients in whom better accuracy is warranted, i.e. when drug dosing is important and especially when renally excreted and potentially toxic drugs and contrast media are administered.
A clinically important finding in this study is that both creatinine- and cystatin C-based equations systematically overestimate kidney function in patients with primary neuromuscular diseases, especially in patients with reduced kidney function. The reason for this overestimation when using cystatin C-based estimates is not entirely clear since previous studies have shown that cystatin C is less correlated to muscle mass and diet than creatinine [
33,
34]. Previously identified non-GFR-related factors such as inflammation, obesity [
17], insulin resistance [
35,
36], oxidative damage [
37], growth hormone [
38], thyroid hormone [
39] and glucocorticoids [
10] have all instead been linked to increased cystatin C production and thus an underestimation of GFR. A potential explanation for the overestimation of kidney function when using cystatin-C based equations could be that body fat is a determinant of cystatin C [
12] and that patients with muscular dystrophy would have not only a reduction in muscle mass but also in fat mass. This could be true but might not be the entire explanation since many patients with muscular dystrophy actually have either an absolute or a relative increase in body fat [
18,
19]. To our knowledge no factor resulting in a true decreased production of cystatin C, and hence in an overestimation of kidney function, has been reported.
The convention of indexing glomerular filtration rate to BSA attempts to normalize kidney function across populations of differing body size but may be inappropriate when a more precise estimation of kidney function is needed or in patients with extreme body sizes. Adding to the complexity, there is a disproportionate relation between extracellular volume (ESV) and BSA [
40], and a higher ECV in women may be concealed by scaling to BSA [
41]. The use of absolute values has been shown to improve the performance of estimation of individual kidney function [
42]. We therefore explored the use of both indexed and unindexed measures and estimates of kidney function in these subjects with muscle wasting diseases. In our study only minor changes were observed when unindexed values were used instead of BSA-indexed ones. One potential reason is that the patient population in this study, despite having muscle wasting conditions, did not display extreme body size. They also had a relatively normal BSA (1.88 m
2) close to the BSA used for normalization (1.73 m
2) of kidney function.
A weakness when using cystatin C-based estimations of kidney function has previously been the lack of an international calibrator resulting in analytical bias and inability to compare cystatin C analysis carried out using different assays. Recently, an international cystatin C calibrator was used to develop an assay-independent cystatin C-based equation for estimation of GFR (CAPA) [
26]. It is however worth pointing out that CAPA does not have a sex coefficient and a potential impact of sex has not been evaluated in this study. In this study cystatin C was calibrated and the matching CAPA equation was used. This has not been done previously in studies with patients with low muscle mass. Interestingly, in this population, a combined equation based on the mean value of creatinine- and cystatin C-based estimates was not more precise and accurate than a cystatin C-based estimation; something that has been shown in other populations [
30]. A plausible explanation for this finding is the overestimation of kidney function for both creatinine- and cystatin C-based estimations. Among the creatinine-based equations evaluated, MDRD had the best accuracy in patients with reduced kidney function, both for P30 and to a lesser extent also P10. MDRD is currently the most commonly used creatinine-based equation in subjects with reduced kidney function.
The findings of our study should however be interpreted in light of its strengths and limitations. The strengths include that the study population is larger than in previous corresponding studies, the measurements of serum creatinine and cystatin C were done using standardized and calibrated assays, and the analysed blood samples were drawn simultaneously with the measurements of iohexol clearance. However, there are also some limitations. One is that we pooled patients with different primary neuromuscular diseases who may have had different disease characteristics, and that from the initial 418 patients identified at the outpatient clinic data on renal function was obtained only in 145 cases in CKD stage I–IIIa, thus few participants had severe kidney failure or an advanced neuromuscular disease with very low muscle mass. The generalizability to all patients with reduced muscle mass should thus be further evaluated. The calibration of measured and estimated kidney function by BSA has been questioned in patients with altered body composition, i.e. both low and high body mass index, however, this concern is not specific for this study [
40,
43] but nonetheless, we have tried to address this issue by calculating both indexed and unindexed measures and estimates.
Another limitation and potential concern is that plasma clearance of iohexol used to measure kidney function in this study has not been validated specifically in patients with primary neuromuscular disease or in other populations with low body mass index and reduced muscle mass, i.e. sarcopenia, but there are studies ongoing to assess this matter (personal communication).
Another potential explanation for part of the results in this study may thus be that both the MDRD and the CKD-EPI equation were developed using urinary clearance of iothalamate as a measurement of kidney function; GFR. This is in contrast with our study but also in the development of the cystatin C -based estimation using CAPA where plasma clearance of iohexol was used. A recent paper has shown that there may be an approx. 15% difference between urinary clearances for the two tracers [
44]. Since plasma clearance is thought to be greater than urinary clearance, plasma clearance for iohexol may be 5–10% lower than urinary clearance for iothalamate, therefore, this must be taken into account.
It should also be highlighted that Pearson's correlation coefficient does not take into account any differences in values between compared groups and thus the correlation can be quite good even when eGFR values differ from measured clearance values by twofold or more. To explain the difference between agreement and correlation it is therefore important to depict bias and the limits of agreement using Bland–Altman diagrams.
In conclusion, our findings indicate that a cystatin C-based estimation of kidney function may be more accurate in patients with primary neuromuscular disease and low muscle mass. However, even though it provides better accuracy, it varies within a wide range and should thus not be considered good enough for use in patients in whom better accuracy is warranted, for example when contrast media and potentially toxic drugs are administered.
Another important finding is that both creatinine- and cystatin c-based equations systematically overestimated kidney function, especially in patients with reduced kidney function. This is clinically relevant, since detecting and diagnosing patients with reduced kidney function is extremely important as they are at risk of developing secondary metabolic complications and/or end-stage kidney disease, and the reduction of kidney function may affect drug dosing. Further studies are necessary both to validate currently used techniques for measuring kidney function in patients with low body mass index and reduced muscle mass, as well as to evaluate diagnostic strategies for estimating kidney function in patients with reduced muscle mass and altered body composition.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.