A Comparative Study on the Efficacy Between Cystatin C and Creatinine-Based Equations for Early Detection of Renal Damage in Patients of Eastern India

Abstract

Chronic kidney disease (CKD) is one of the leading causes of mortality across the globe. Early diagnosis of the disease is important in order to prevent the adverse outcome related to CKD. Many laboratories adopt creatinine-based e-GFR equations which yields imprecise results leading to misdiagnosis of CKD. Emerging studies indicated cystatin C as a better renal marker than creatinine. The aim of the study is to compare the efficacy of CKD epidemiology collaboration (CKD-EPI) creatinine e-GFR equations with (CKD EPI) cystatin-based e-GFR equations alone and in combination with creatinine for early detection of CKD. A cross-sectional study employing 473 patients was conducted. Three estimating GFR equations were calculated based on creatinine and cystatin C. Pearson Correlation study was done to assess the correlation of creatinine and cystatin C with their respective GFRs. A predictive model was developed, and ROC curve was constructed to compare efficacy, sensitivity and specificity of the creatinine and cystatin C based equations. Cystatin C exhibited better negative correlation with GFR than creatinine in correlation study performed with three commonly employed eGFR equations including  CKD EPI Creatine cystatin C combined  equation (2021), cys C alone and CKD EPI  creatinine (2021)  equations respectively[r=(–) 0.801 vs. r=(–)0.786 vs. r=(–)0.773]. Predictive model demonstrated highest efficiency, sensitivity and specificity for creatinine-cystatin C combined equation (88%, 81% and 93%) followed by cystatin C alone equation (73%,63% and 82%) and creatinine-based equation  (61%, 56% and 66% respectively). The study showed better performance of cystatin C based equations for early detection of advance stages in chronic kidney disease as compared to creatinine-based e-GFR equation.

Introduction

CKD has emerged as a leading health problem across the world [1–4] 0.37 million of American adults are reported to have CKD [5]. CKD is not restricted to the older population, but it is equally prevalent in younger generation [6]. In last few decades there is also an enormous surge in CKD in Indian population [7].Diabetes and hypertension are mentioned as the most important contributors to the underlying pathology of kidney disease [8, 9].Existing reports indicated that 40–60% of the CKD population irrespective of urban or rural areas had diabetes and hypertension. Synergistic interaction between these ailments increases the chances of CKD [10].Early diagnosis of disease is of paramount importance as patients with CKD remain asymptomatic or show negligible symptoms at the onset of the disease, more importantly they are many times more vulnerable to cardiovascular outcomes which are reported to claim more lives than end stage renal disease (ESRD), an advanced progression of kidney disease in 15–20 years [11].

Estimated glomerular filtration rate (e-GFR) is a rapid, simple and authentic way to measure declining kidney function in the patients who are high candidates to have CKD. Several internationally adopted e-GFR equations are used in ambulatory settings to evaluate decreased kidney functions. e-GFR formulae has gained popularity over 24 h creatinine clearance and radioactive Tc-99 m DTPA (diethylene-triamine-pentaacetate) to measure GFR globally [12], the gold standard method to measure GFR but the procedures are hazardous, elaborative and cumbersome to carry out on a regular basis [13].

Epidemiological studies highlighted the limitations of using creatinine-based equations to measure GFR owing to its imprecise estimation leading to over diagnosis of CKD [14–16].With the growing research on kidney biomarkers cystatin C (cyc C), a serine protease secreted by all nucleated cells, has been recognized as a promising renal marker which unlike creatinine is dependent on the variables like age, sex, muscle mass and diet [17, 18]. CKD epidemiology collaboration (CKD-EPI), a collaborative group showcased the advantage of cysC over creatinine as an endogenous marker and developed two modified equations named CKD–EPI-cys C (2012) and CKD-EPI-cys C in combination with creatinine (2021) for accurate prediction of CKD [19]. Inker et al. developed new creatinine and cys C based equations without involving race in 2021 [20]. Accumulated evidence suggested the usefulness of cys C based equations for better cardiovascular risk stratification in the patients with CKD [21–23]. There is a real dearth of data on using cys C based e-GFR equations for diagnosis of CKD in the population of Eastern India. The objective of the study is to assess decline in renal function in CKD patients of Eastern India using modified E-GFR equations without race and to compare the performance of the creatinine and cystatin C based equations for detecting early renal impairment in them in order to delay the adverse outcome of chronic kidney disease.

Materials and Methods

A total of four hundred seventy-three (n = 473) patients were  recruited for the study after initial screening based on inclusion and exclusion criteria from the Nephrology department of Ramakrishna Mission Seva Pratishthan (RKMSP) hospital, Kolkata, India over a period of 3 years All the patients consented for the study and a detailed questionnaire including anthropological parameters, existing illness, history of other co morbidities, details of present and past medications were filled by each patient before commencement of the study. Institutional Ethics Committee (IEC) clearance was obtained for the study according to the protocol Serum creatinine (cr) and cystatin C were tested for all participants by enzymatic method traceable to IDMS (Isotope Dilution Mass Spectometry) and immune nephelometric methods traceable to international calibrator standard ERM-DA471/IFCC employing Vitros 4600 (Ortho Clinical Diagnostics, Germany) and Mispa i3 (Agape Diagnostics, India) autoanalyzers respectively. CKD-EPI-creatinine (2021), CKD-EPI-cys C (2012) and CKD-EPI-cr-cys C (2021) combined equations (e-GFR) were used to estimate GFR in the enrolled patients. CKD is defined as kidney damage or glomerular filtration rate (GFR) < 60ml/min/1.73 m² for 3 months or more irrespective of causes (KDIGO: Kidney Disease Improving Global Outcome) and is classified based on cause, GFR category and albuminuria category (CGA classification) [24]. GFR categories of CKD is as follows:

Grade I (GI): minimal damage with normal GFR: ≥90 ml/min; GII: mild damage with slightly decreased GFR: 60–89ml/min, GIIIa mild to moderately decreased GFR: 45–59 ml/min, GIIIb: moderately to severely decreased GFR: 30–44ml/min, GIV: severely decreased GFR: 15–29 ml/min and GV: kidney failure, GFR :< 15 ml/min. In 2015 International Classification of disease (ICD)-CM-10 used codes N18 for staging CKD (N18.1CKD, stage I,N18.2CKD stage II,N18.3CKD stage III,N18.4CKD stage IV,N18.5CKD stage V.ICD-10-CM coding CKD stage III (N18.3) has been revised and G3 stratified into N18.31 CKD, Stage IIIa (GFR = 45–59) and N18.32 CKD, Stage IIIb (GFR = 30–44).

Statistical Analysis

All data were expressed in mean ± SD. A p value < 0.05 was considered as statistically significant. SPSS version 10, software was used for statistical analysis. Pearson Correlation study was performed for three e-GFR equations. Hosmer and Lemeshow goodness of fit (GOF)l test was done to evaluate the best fit equation to predict early renal outcome. AUC (area under curve) value in ROC (receiver operator curve) analysis measures entire two-dimensional value in the curve. AUC value 0.7–0.8 considered acceptable, 0.8–0.9 considered excellent and considered to be 100% accurate when it reaches to 1.0. A predictive model was constructed to compare efficacy, sensitivity and specificity among three equations for disease diagnosis.In predictive model deviance defines to the difference of likelihood between a saturated model and the fitted model. Lower value of deviance shows better fit whereas higher value corresponds to a less accurate model. Similarly, AIC (Akaike Information Criterion) value estimates the prediction error for a given set of data, hence accuracy of the proposed model can be judged. Lower AIC value demonstrated best fit model.

Results

There are 278 males and 195 females in a whole of 473 study participants. Demographic and laboratory measurements of the patients are registered in Table 1. In the present cohort of CKD patients, diabetes mellitus and hypertension attribute to the development of the disease in 39.7% (188 out of 473 participants) and 52.6% (249 out of 473 participants)respectively whereas other diseases accounts for CKD in 7.6% (36 out of 473) patients. We found a stronger negative correlation of estimated GFR with cystatin C in comparison to creatinine (Table 2). A significant gender disparity was observed in the prevalence of CKD in our study. CKD was found to be more prevalent in males than their female counterparts (Table 3). The study revealed  cystatin C based equations were  more effective to diagnose CKD with decline in renal function (stage III to stage V) as compared to creatinine-based equations (Table 4). A predictive model was developed for comparing the performance of three established equations: CKD-EPI-cr (2021), CKD-EPI-cys C (2012) and CKD-EPI-cr-cys C (2021) for early detection of CKD. Least deviation of observed values from predicted ones is observed by their residual deviance (AIC). Creatinine-cystatin C combined equation (CKD-EPI-2021) showed its superiority in performance over other two equations (cystatin C alone and CKD-EPI-creatinine) (Table 5). Hosmer and Lemeshow goodness of fit (GOF)test was conducted to evaluate the best fit equation to predict early renal impairment and it demonstrated the superior performance of e-GFR based on cysC in comparison to creatinine (Table 6).

Table 1.

Demographic and laboratory parameters of CKD patients

Parameters	Mean ± SD
Age (years)	52.35 ± 14.50
Height (cm)	159.29 ± 9.07
Weight (kg)	61.11 ± 11.70
Body mass index (kg/m2)	24.01 ± 4.59
Systolic pressure (mm of Hg)	130.85 ± 17.07
Diastolic Pressure (mm of Hg)	80.59 ± 10.02
Serum creatinine (mg/dl)	1.31 ± 0.66
Serum cystatin C (mg/l)	1.36 ± 0.91
e-GFR CKD-EPI-Cr (ml/min/1.73m2)	70.32 ± 28.96
e-GFR CKD-EPI-Cys C (ml/min/1.73m2)	73.30 ± 39.15
eGFR CKD-EPI-Cr-Cys C (ml/min/1.73m2)	72.58 ± 33.45

Table 2.

Correlation between creatinine and/or cystatin C with three commonly employed e-GFR equations

Parameter	Equation applied	r value	p value
Cr vs. e-GFR	CKD-EPI-cr(2021)	− 0.773	< 0.01
Cys C vs. e-GFR	CKD-EPI-cysC (2012)	− 0.786	< 0.01
CysC vs. e-GFR	CKD-EPI-cr-cysC (2021)	− 0.801	< 0.01

Table 3.

Comparison of e-GFR in males and females among different equations

e-GFR	CKD-EPI-cr				CKD-EPI-cys C				CKD-EPI-cr–cys C
ml/min/1.73m2	F (n)	%	M (n)	%	F (n)	%	M (n)	%	F (n)	%	M (n)	%
GI	72	37	65	23	79	40	94	34	70	36	85	31
GII	50	26	93	33	44	23	58	21	59	30	77	28
GIIIa	24	12	70	25	18	9	39	14	18	9	52	19
GIIIb	33	12	35	17	17	9	48	17	20	10	40	14
GIV	12	6	13	5	28	14	32	11	23	12	20	7
GV	4	2	2	1	9	5	7	3	5	3	4	1

Table 4.

Reclassification of CKD patients using cys C based e-GFR equations

Table 5.

Comparison of efficacy amongst various e-GFR equations in diagnosis of CKD

Predictive model	Deviance at 496 degrees of freedom	AIC value	Efficiency (%)	Sensitivity (%)	Specificity (%)
CKD-EPI-cr(2021)	659.19	667.19	0.61	0.559	0.661
CKD-EPI-cys C (2012)	549.30	557.3	0.726	0.631	0.818
CKD-EPI-cr-cysC (2021)	259.75	269.75	0.878	0.813	0.930

Table 6.

Hosmer and Lemeshow goodness of fit (GOF)l tes

Equation	AUC values
CKD-EPI-cr (2021)	0.639
CKD-EPI-cystatin C (2012)	0.787
CKD-EPI-creatinine-cystatin C (2021)	0.930

cystatin C showed best correlation with e-GFR [r=(−) 0.801] in CKD EPI cr-cysC combined equation than cys C alone and creatinine based e-GFR used to detect renal impairment.

CKD was found to be present in males predominantly in stage III a and III b as compared to their female counterparts using cys C based equation in this study.

With decrease in GFR Cys C based equation found to be more effective for disease diagnosis as compared to creatinine based equation.

(Colour code was used to indicate severity of the condition. While green expresses minimum level of disease severity, red expresses maximum level of disease severity)

In this study predictive model constructed for all three equations. The best model was observed for cysC_cr combined equation (AIC value = 269.75) followed by cystatin C equation ((AIC value = 557.3) and cr based equation (AIC value = 667.19) respectively.

High AUC value (0.930) for cysC_cr  equation than other two equations (AUC for cys C equation = 0.787 and AUC for creatinine based equation = 0.639) demonstrated high sensitivity and specificity of the equation for detection the disease (Fig. 1).

Fig. 1.

ROC curve for different e-GFR equations

The extreme right box with green colour denotes ROC curve for CKD EPI cr-cys C equation, the middle box with red colour denotes ROC curve for CKD EPI cys C equation and extreme left box with the blue colour shows ROC curve for CKD EPI creatinine equation.

Discussion

The present study demonstrated the necessity of employing cys C based equation for early diagnosis and accurate classification of chronic kidney disease. The roles of creatinine –based equations have been questioned for ages in early disease detection. In general creatinine-based equations underestimate the disease prevalence in the initial stages of CKD as creatinine remains within reference range till 50% of the glomerular function is lost (e-GFR: 60–80 ml/min/1.73 m² considered as creatinine blind zone) in comparison to cys C based estimation of GFR [25].Recent studies indicated that a significant percentage of global population including India still remains undetected at advanced stages by adopting e-GFR equations (like MDRD, CG and CKD EPI creatinine e-GFR equations)which use creatinine as an endogenous marker as they overestimate GFR in the patients with GFR ≤ 60 [14].The importance of utilizing cyst C based equation lies there for early detection of reduced kidney function especially for high-risk group patients, most likely patients with diabetes and hypertension who live with a false sense of relief of not having CKD while actually harboring the disease in them as CKD remains asymptomatic in initial stages. Our earlier study showed the effectiveness of cys C as a biomarker for early disease diagnosis in the patients with hypertensive nephropathy [26]. On several occasions CKD is diagnosed in such an advanced stage, more often CKD stage III when there is no other option left but to dialysis, this could be avoided if timely clinical intervention is taken.

Gender related inequities are well documented in numerous studies as men and women differ in biological susceptibility to several diseases including CKD [27] Gender disparity is also reflected in CKD related mortalities [28]. Global prevalence of CKD is more pronounced in female whereas ESRD (end stage renal disease) is reported to be higher in males [29]. Increased prevalence of obesity and hypertension in females might account for Female predominance in Chronic Kidney disease but true reason behind the difference is yet to clear. Reports highlighted faster progression of CKD in males due to high level of testosterone, local made alcohol consumption, exposure to pesticide during their field work etc.; hence kidney replacement therapy and dialysis are found to be higher in men than women. Unequal access to treatment results in sex disparity should also be taken in to account. It is interesting to note that hospital based studies indicated male predominance in CKD which is in accordance with our study findings. In this present study male predominance of CKD was observed in stage IIIa and IIIb. CKD-EPI-Cys c equation detected 14% male versus 9% female in CKD stage IIIa and 17% male versus 9% female in CKD IIIb where as CDK-Cys c combined equation demonstrated 19% male versus 9% female in CKD stage IIIa and 14% male versus 10% female in CKD stage IIIb. Similar trends of results were observed by using CKD-EPI-Creatinine based equation (Table 3) There might be several social and structural reasons behind this inversion of sex ratio in hospital set up. Men have easier access to expensive treatment due to the ability to meet up financial requirements on the contrary women is deprived of medical attention due to their inferior socio economic condition and inability to pay hospital bills.

Numerous studies demonstrated the better performance of cys C based eGFR equations over creatinine based e GFRs in predicting kidney disease when GFR is ≤ 60 [30, 31] as muscle mass, extra renal elimination and tubular secretion influence creatinine in measuring correct GFR.Our results are compatible with other studies conducted around the globe [32, 33]. Cys C based equation has received much attention due to its superiority to assess disease progression. In our study we observed best reciprocal correlation of cys C with e-GFR in CKD EPI cr_cys C combined equation (r=-0.801) followed by cys C alone (r=−0.786) and CKD EPI creatinine e GFR equations (r= −0.773) (Table 5). Highest AUC value ( AUC=0.930) was observed for CKD EPI  Creatine and cys C combined equation  than  CKD EPI cystatin C  equation  alone (AUC = 0.787)  and  CKD EPI creatinine (AUC = 0.639) eGFR equations. We also constructed a predictive model for comparing efficacy of the three equations which showed highest sensitivity (81%) and specificity (93%)for Cys C-cr combined equation than cys C alone equation (sensitivity 63% and specificity 82%) followed by creatinine-based equation (sensitivity 56% and specificity 66%) for disease diagnosis. We reclassified 473 patients using CKLD-EPI-cyc C based equations who were prior classified into stages of CKD (GI-GV) based on CKD-EPI (cr) 2021 equation. Cys C-eGFR equation alone and in combination with creatinine detected 43 (9%) and 60 (13%) patients in stage IV followed by 9 patients (2%) and 16 patients (3%) in stage V respectively whereas creatinine based equation diagnosed only 25 patients (5%) in stage IV and 6 patients (1%) stage V respectively indicating the better performance of cys-C-eGFR over cr-based eGFR for diagnosing disease at advanced stages.It is disappointing to see that cysC still battles for it’s space in the arena of nephrology despite being a superior renal marker than creatinine due to unaffordability for the users. It is six times expensive than creatinine but we feel cost of cystatin C is justified as it is helpful for making  early and accurate evaluation of kidney disease and offer timely clinical intervention.

Conclusion

Cystatin C based equation should be opted for detection of early renal impairment in the patients with advanced staged chronic kidney disease in Eastern India. Hospital based study findings might not be a true reflection of the status of disease in the population. A population based study with larger participants is needed to validate our study findings.

Limitations

1. We could not measure GFR by radioactive Tc-99 m DTPA (diethylene-triamine-pentaacetate) method as we do not have permissible settings in our hospital to conduct the test.

• (B)Proteinuria and ACR (microalbumin creatinine ratio) could not be measured in CKD patients as most of the patients did not turn up for urine testing.

Funding

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Declarations

Conflict of interest

The authors declare no conflict of interest for this study.