Shrunken pore syndrome and mortality: a cohort study of patients with measured GFR and known comorbidities

Abstract

Shrunken pore syndrome (SPS) is defined by a cystatin C-based estimation of glomerular filtration rate (eGFR_CYS) being less than 60% or 70% of a creatinine-based GFR estimation (eGFR_CR) in the absence of extrarenal influences on cystatin C or creatinine concentrations. SPS has been associated with a substantial increase in mortality or morbidity in all investigated populations. However, in these studies, neither the diagnoses, nor causes of death were described, and only estimated GFR was available. The present study concerns 2781 individuals with measured GFR (mGFR), known diagnoses, and known causes of death during 5.6 years in median. Cox multivariate proportional hazards regression model was used to estimate hazard ratios (HR) for all-cause and cancer, cardiovascular, diabetes or chronic kidney disease (CKD) as cause-specific mortality among patients with SPS. At an eGFR_CYS/eGFR_CR-ratio <0.70, the adjusted SPS death risk in the total cohort (HR 3.0, 95% CI 2.4–3.7) was clearly higher than that for the other diagnosis groups. In a sub-cohort of 1300 persons with or without diagnosis, but with normal mGFR, the all-cause mortality of SPS was markedly increased (HR 4.1, 95% CI 2.6–6.5). In a sub-cohort of 567 persons with normal mGFR and no diagnosis, the all-cause mortality of SPS was even more increased (HR 7.3, 95% CI 2.3–23). The prevalence of SPS in the total cohort was 23% and in the sub-cohorts 17 and 12%, respectively. As SPS is associated with a high mortality, occurs in the absence of reduced mGFR and albuminuria, it expands the spectrum of kidney disorders.

Keywords:

• Creatinine
• cystatin C
• GFR
• kidney
• mortality

Introduction

According to studies by Norden et al. the glomerular sieving coefficients for 10–30 kDa proteins in the human kidney are relatively high with coefficients between 0.07 and 0.9 [1]. Similar sieving coefficients for proteins of corresponding sizes have been observed in the rat kidney [2]. These relatively high sieving coefficients, combined with the high production of ultrafiltrate in health, means that peptides and proteins ≤30 kDa in plasma normally are mainly catabolised by the kidneys and have a high turnover. For example, at least 85% of the catabolism of cystatin C occurs in the kidney [3]. The production of ultrafiltrate corresponds to the glomerular filtration rate (GFR), which usually is measured, or estimated, using molecules like creatinine, ⁵¹Cr-EDTA, iohexol, or iothalamate with molecular masses between 0.113 and 0.821 kDa. A pertinent question is then whether a reduction in the renal excretion of these small molecules always will correspond to an exactly equal reduction in the renal excretion of 10–30 kDa molecules. Invasive clearance studies in humans and rats show that this is not the case [4–8]. In one of these studies, neutral dextrans of various sizes were used in pregnant females and showed, that the renal clearance of molecules corresponding 10–30 kDa proteins, is decreased in the third trimester, whereas the renal excretion of small molecules, corresponding to measured GFR (mGFR), is not [6]. In another study of patients with diabetic nephropathy, neutral dextrans or cystatin C were used to measure glomerular filtration of molecules of various sizes, and it demonstrated that the filtration of cystatin C was reduced before that of iothalamate due to a reduction in mean glomerular pore size [7]. We have interpreted our observations of a selective accumulation of 10–30 kDa proteins in the third trimester of pregnancy despite a normal GFR [9,10], as consequences of the selective reduction in the renal clearance of 10–30 kDa molecules shown by previous invasive studies [6] and corresponding to shrunken pores in the pore model of glomerular filtration [11,12]. We have suggested that a simple way of demonstrating such a size-selective reduction in the glomerular filtration rate is to compare the ratio of a cystatin C- (eGFR_CYS) and a creatinine (eGFR_CR) based estimate of GFR [11]. If the eGFR_CYS/eGFR_CR-ratio is <0.60 or <0.70, in the absence of known extrarenal influences on the levels of cystatin C or creatinine, we have suggested that it represents the presence of a syndrome tentatively called shrunken pore syndrome (SPS) [11]. Several studies of SPS have been done and all of them show that the long-term mortality [13–17], or morbidity [17,18], of individuals with SPS are strongly increased, even in the absence of reduced GFR.

But in all these studies, neither the diagnoses present when SPS was investigated, nor the death causes were reported, and no data for measured GFR available. The aim of the present study, which was based on the Lund Cystatin C Standardization (LCS) cohort with known mGFR [19], was to study the association between SPS and mortality while controlling for general patient characteristics and disease history.

Material and methods

Study cohort

The cohort represents the patients’ first GFR examination in the Lund Cystatin C Standardization (LCS) cohort, primarily established to generate a new cystatin C-based GFR-estimating equation (the CAPA equation) [19]. The LCS cohort is based on consecutive Swedish Caucasian (≥99%) patients above 18 years referred for determination of GFR at Skåne University Hospital, Lund, from May 2008 to March 2010. Common causes for referral were manifest or suspected diabetic nephropathy, interstitial nephritis, glomerulonephritis, nephrotic syndrome, haematuria, proteinuria, reflux nephropathy, myeloma, vasculitis, consideration of initiation of haemodialysis, evaluation of potential renal donors, control after kidney transplantation, and to dose drugs cleared by the kidneys. Among the original 2847 cohort members, 42 were excluded due to invalid personal registration number and 24 due to survival less than one month, leaving 2781 unique patients for the present study (Table 1). The study was approved by the local ethics committee (permissions LU 2015/860 and 2016/169). Patient data and samples were treated pseudonymized in all statistical analyses.

Table 1. Basic characteristics of the Lund Cystatin C standardization cohort, stratified by the quartiles of the ratio of estimated GFR according to CAPA_CYS and LMR_CR.

	Total	CAPACYS / LMRCR ratio	CAPACYS / LMRCR ratio	CAPACYS / LMRCR ratio	CAPACYS / LMRCR ratio
		<0.70	0.70–0.84	0.85–0.99	≥1.00
N	2781	645	699	728	709
Age, years	63 (20–85)	64 (24–86)	63 (22–87)	62 (20–84)	62 (19–85)
Females, % (n)	48.6 (1 332)	43.4 (280)	47.5 (332)	51.0 (372)	51.9 (368)
BMI, kg/m^2	26 (18–39)	26 (18–40)	27 (19–40)	26 (18–37)	26 (19–36)
Heigth, cm	170 (152–190)	171 (152–190)	170 (152–190)	170 (151–191)	170 (154–190)
Weight, kg	76 (49–115)	77 (46–119)	78 (50–121)	75 (49–112)	74 (50–109)
BSA^a	1.9 (1.5–2.3)	1.9 (1.4–2.4)	1.9 (1.5–2.3)	1.9 (1.4–2.3)	1.9 (1.5–2.3)
LBM^b	53.6 (37.5–75.3)	54.2 (36.2–76.0)	54.3 (38.2–75.5)	53.3 (37.0–74.7)	52.4 (38.9–75.7)
mGFR^c	57 (10–118)	46 (8–106)	56 (10–118)	66 (12–114)	59 (10–123)
Cystatin C, mg/L	1.45 (0.75–4.38)	1.94 (1.07–5.30)	1.41 (0.88–4.33)	1.17 (0.78–3.70)	1.20 (0.67–3.92)
Creatinine, µmol/L	95 (45–427)	98 (43–422)	92 (44–388)	88 (46–356)	109 (47–499)
CAPACYS^c	47 (10–107)	33 (7–71)	49 (10–91)	61 (13–101)	59 (11–124)
LMRCR^c	61 (11–111)	59 (12–116)	63 (12–114)	66 (13–108)	53 (10–106)
CAPACYS / LMRCR ratio	0.86 (0.46–1.29)	0.61 (0.34–0.70)	0.78 (0.70–0.85)	0.92 (0.85–0.99)	1.10 (1.00–1.44)
CKD-EPICYS^c	46 (10–112)	31 (8–74)	48 (10–97)	60 (12–109)	58 (12–119)
CKD-EPICR^c	66 (11–127)	63 (12–129)	68 (12–130)	72 (13–126)	56 (8–125)
CKD-EPICYS / CKD-EPICR ratio	0.78 (0.41–1.26)	0.54 (0.32–0.81)	0.70 (0.58–0.93)	0.84 (0.70–1.00)	1.00 (0.84–1.51)
FASCYS / FASCR ratio	0.84 (0.47–1.34)	0.62 (0.39–0.96)	0.77 (0.59–1.09)	0.89 (0.70–1.21)	1.05 (0.83–1.167)
FASCYS	48 (15–109)	36 (12–79)	49 (15–96)	58 (18–105)	59 (17–128)
FASCR	62 (13–140)	60 (14–147)	64 (14–145)	67 (15–131)	55 (11–128)
Cancer, % (n)	28.6 (794)	28.5 (184)	30.8 (215)	29.3 (213)	25.7 (182)
CVD, % (n)	27.0 (750)	34.1 (220)	28.0 (196)	22.9 (167)	23.6 (167)
Diabetes, % (n)	11.1 (308)	17.8 (115)	12.7 (89)	8.2 (60)	6.2 (44)
CKD, % (n)	14.8 (411)	18.0 (116)	13.7 (96)	10.9 (79)	16.9 (120)

^aBody surface area, calculated using Dubois formula 0.007184 × weight^0.425 × height^0.725.

^bLean body mass, calculated using James formula for women as 1.07 × weight – 148 × (weight/height)² and men as 1.10 × weight – 128 × (weight/height)². Weight in kg and height in cm.

^cmL/min/1.73 m². Unless otherwise noted, continuous characteristics are presented as median (2.5–97.5 percentiles); categorical values are presented as percent (count).

Measurement of GFR

GFR was measured as plasma clearance of iohexol [20–22].

Measurements of cystatin C and creatinine

Cystatin C was determined by a particle-enhanced immunoturbidimetric method using a reference material traceable to the international cystatin C calibrator [19,23] and creatinine by an enzymatic colorimetric assay using a calibrator traceable to a primary reference material with values assigned by isotope dilution mass spectrometry [24].

Register linkage

The cohort was followed longitudinally from one month after the date of the first GFR determination during 2008–2010 with respect to survival in the Swedish population register until Feb 28, 2016.

Data on cause-specific deaths were obtained from the cause of death register hosted by the National Board of Health and Welfare in Sweden. The cause-specific deaths were grouped according to ICD-10 as cancer (C00.0–C97.9), CVD (I10.0–I82.9), DM (E10.0–E14.9), or CKD (N18.1–N18.9). Each patient´s history of disease was obtained from the patient register and grouped in the same way as cause-specific deaths. Cancer and CVD diagnoses were included from 5 years before, DM and CKD included all previous diagnoses and up to 1 month after the date of the patients’ first GFR determination. The date of diagnosis closest to the date of the patients’ first GFR determination was used.

Statistical analysis

All statistical analyses were conducted using STATA version 14.2 (StataCorp.). Kaplan–Meier estimates were used to present sex-specific survival probabilities in four eGFR_CYS/eGFR_CR-ratio intervals using the cystatin C-based CAPA-equation [19] (CAPA_CYS) and the creatinine-based LMR equation (LMR_CR) [25]. Cox regression was used to model the association between the categorized eGFR_CYS/eGFR_CR-ratio and all-cause and cause-specific mortality with adjustment for age (continuous; per 10 years), sex, BMI (<20, 20–24, 25–29 or ≥30 kg/m²), history of disease at baseline (cancer, CVD, DM and CKD), mGFR, eGFR_CR and eGFR_CYS. The GFR-variables were entered categorized as <30, 30–59, 60–89 or ≥90 mL/min/1.73 m² in the regression models. We conducted statistical trend tests by using the eGFR_CYS/eGFR_CR-ratio as a continuous variable in separate analyses. Results were presented as hazard ratios with 95% confidence intervals (CIs), combined or stratified by sex. Several sensitivity analyses were performed with all-cause mortality as outcome: 1) additional categorization of the eGFR_CYS/eGFR_CR-ratio in six groups, 2) estimating the eGFR_CYS/eGFR_CR-ratio with two additional pairs of GFR-estimating equations, 3) restriction to patients without any of the disorders studied (cancer, CVD, DM, CKD; n = 1049), 4) restriction to patients with measured GFR ≥60 mL/min/1.73m² (n = 1300), 5) restriction to patients with both normal measured GFR (≥60 mL/min/1.73m²) and no diagnosis (n = 567), 6) excluding variables (CAPA_CYS, LMR_CR, or mGFR) from the regression models, either one at a time or simultaneously.

Results

All-cause mortality in the total cohort

The study cohort constituted 2781 adult persons consecutively referred to Skåne University Hospital, Lund, for determination of mGFR by iohexol clearance 2008–2010 and followed until 2016 (median follow-up 5.6 years). The distribution of the eGFR_CYS/eGFR_CR-ratio is shown in Figure 1. Table 1 summarizes available baseline characteristics, stratified in four groups based on the eGFR_CYS/eGFR_CR-ratio obtained using the cystatin C-based CAPA-equation (CAPA_CYS) [19] and the creatinine-based LMR equation (LMR_CR) [25]. The mortality in the total cohort was 37.7%. Figure 2 shows the survival stratified by sex. Cox regression analysis was used to study all-cause mortality as a function of different CAPA_CYS/LMR_CR-ratio intervals. There was a significant increase in all-cause mortality for all eGFR_CYS/eGFR_CR-ratio intervals below 1.00 (p_trend <.001; Table 2). The group with the lowest ratio interval (<0.70) displayed an adjusted HR for mortality of 3.0 (95% CI 2.4–3.7) when the ratio interval ≥1.0 was used as reference group. Diagnoses of cancer, diabetes mellitus (DM), cardiovascular disease (CVD), or CKD, all displayed a lower adjusted HR for mortality than SPS defined by an eGFR_CYS/eGFR_CR-ratio of <0.70 (Table 2). Low eGFR based upon cystatin C (CAPA_CYS) was also associated with an increase in mortality, but the HR was lower than that of the SPS-groups defined by the eGFR_CYS/eGFR_CR-ratio intervals <0.70 or 0.70–0.84. Neither a low eGFR (<30mL/min/1.73m²) based upon creatinine (LMR_CR), nor a low mGFR (<30mL/min/1.73m²) was associated with an increase in mortality in the adjusted analysis (Table 2, columns B and C). Associations between low eGFR_CYS/eGFR_CR-ratios and all-cause mortality were present both among male and females (Table 3).

Figure 1. Distribution of the eGFR_CYS/eGFR_CR-ratio, calculated using three pairs of eGFR equations. CAPA_CYS and LMR_CR [19,25]. CKD-EPI_CYS and CKD-EPI_CR [26]. FAS_CYS and FAS_CR [27].

Figure 2. Survival stratified on the quartiles of the ratio of estimated GFR using the CAPA_CYS and LMR_CR equations (<0.70, 0.70–0.84, 0.85–0.99 and ≥ 1.00) for all 1332 female (A) and 1449 male (B) patients.

Table 2. Cox regression for all-cause mortality in the Lund Cystatin C standardization cohort (n = 2781).

	A		B		C
	HR (95 % CI)	p	HR (95 % CI)	p	HR (95 % CI)	p
CAPACYS / LMRCR ratio		<0.001^a		<0.001^a		<0.001^a
<0.70	3.0 (2.5–3.6)		3.2 (2.6–3.8)		3.0 (2.4–3.7)
0.70–0.84	2.2 (1.8–2.7)		2.3 (1.9–2.7)		2.2 (1.8–2.7)
0.85–0.99	1.3 (1.1–1.6)		1.3 (1.1–1.6)		1.3 (1.1–1.6)
≥1.00	1.0 (Ref.)				1.0 (Ref.)
CAPACYS, mL/min/1.73 m^2		<0.001				0.002
< 30	4.1 (3.0–5.7)		–		1.7 (1.0–2.8)
30–59	2.0 (1.5–2.8)		–		1.0 (0.7–1.6)
60–89	1.7 (1.2–2.3)		–		1.1 (0.7–1.6)
≥90	1.0 (Ref.)		–		1.0 (Ref.)
LMRCR, mL/min/1.73 m^2		<0.001				<0.001
<30	2.3 (1.9–2.8)		–		0.7 (0.4–1.1)
30–59	1.2 (1.0–1.5)		–		0.5 (0.4–0.7)
60–89	1.2 (1.0–1.5)		–		0.6 (0.5–0.8)
≥90	1.0 (Ref.)		–		1.0 (Ref.)
mGFR, mL/min/1.73 m^2		<0.001		<0.001		0.43
<30	2.9 (2.4–3.6)		1.2 (0.9–1.5)		1.0 (0.7–1.6)
30–59	1.5 (1.2–1.8)		0.6 (0.5–0.8)		0.9 (0.6–1.2)
60–89	1.5 (1.2–1.9)		0.7 (0.6–0.9)		0.9 (0.7–1.2)
≥90	1.0 (Ref.)		1.0 (Ref.)		1.0 (Ref.)
BMI, kg/m^2		0.007		<0.001		<0.001
<20	1.2 (0.9–1.5)		1.2 (1.0–1.6)		1.2 (1.0–1.5)
20–24	1.0 (Ref.)		1.0 (Ref.)		1.0 (Ref.)
25–29	0.8 (0.7–1.0)		0.8 (0.7–1.0)		0.8 (0.7–1.0)
≥30	0.9 (0.7–1.0)		0.7 (0.6–0.8)		0.7 (0.6–0.8)
Age, per 10 years	1.65 (1.57–1.73)	<0.001	1.57 (1.48–1.66)	<0.001	1.58 (1.49–1.68)	<0.001
Female	1.2 (1.1–1.4)	0.003	1.1 (1.0–1.3)	0.051	1.1 (1.0–1.3)	0.042
Cancer	2.6 (2.3–2.9)	<0.001	2.4 (2.1–2.7)	<0.001	2.5 (2.2–2.9)	<0.001
CVD	1.4 (1.2–1.6)	<0.001	1.1 (1.0–1.3)	0.087	1.1 (1.0–1.3)	0.076
Diabetes	1.6 (1.3–1.8)	<0.001	1.2 (1.0–1.5)	0.024	1.2 (1.0–1.5)	0.024
CKD	1.3 (1.2–1.6)	<0.001	1.1 (0.9–1.3)	0.28	1.1 (0.9–1.3)	0.50

A: unadjusted (univariate) analysis evaluating each variable at a time.

B: adjusted model including all variables in the table, except CAPA_CYS and LMR_CR.

C: adjusted model including all variables in the table.

^a p-value for trend, from separate analysis using the CAPA_CYS / LMR_CR ratio as a continuous variable rather than as categorical.

Table 3. Cox regression for all-cause mortality in the Lund Cystatin C standardization cohort stratified by sex (males; n = 1429, and females; n = 1352)

	Male		Female
	HR (95 % CI)	p	HR (95 % CI)	p
CAPACYS / LMRCR ratio		<0.001^a		<0.001^a
< 0.70	2.6 (1.9–3.6)		3.2 (2.3–4.4)
0.70–0.84	1.9 (1.4–2.6)		2.5 (1.8–3.3)
0.85–0.99	1.0 (0.7–1.4)		1.5 (1.2–2.0)
≥ 1.00	1.0 (Ref.)		1.0 (Ref.)
CAPACYS, mL/min/1.73 m^2		0.010		0.15
< 30	2.6 (1.1–6.0)		1.3 (0.6–2.7)
30–59	1.5 (0.7–3.3)		0.9 (0.5–1.5)
60–89	1.3 (0.6–2.6)		1.0 (0.6–1.6)
≥90	1.0 (Ref.)		1.0 (Ref.)
LMRCR, mL/min/1.73 m^2		<0.001		0.018
<30	0.4 (0.2–0.9)		1.0 (0.6–2.0)
30–59	0.4 (0.2–0.7)		0.7 (0.4–1.1)
60–89	0.6 (0.4–0.9)		0.7 (0.5–1.0)
≥90	1.0 (Ref.)		1.0 (Ref.)
mGFR, mL/min/1.73 m^2		0.40		0.74
<30	1.0 (0.5–1.9)		1.0 (0.5–1.7)
30–59	0.8 (0.4–1.4)		0.9 (0.5–1.4)
60–89	0.8 (0.5–1.3)		1.0 (0.7–1.5)
≥90	1.0 (Ref.)		1.0 (Ref.)
BMI, kg/m^2		0.002		<0.001
<20	1.1 (0.7–1.7)		1.3 (1.0–1.7)
20–24	1.0 (Ref.)		1.0 (Ref.)
25–29	0.7 (0.6–0.9)		0.9 (0.7–1.1)
≥30	0.7 (0.5–0.9)		0.7 (0.5–0.9)
Age, per 10 years	1.72 (1.57–1.89)	<0.001	1.47 (1.36–1.59)	<0.001
Cancer	2.4 (2.0–2.9)	<0.001	2.6 (2.2–3.2)	<0.001
CVD	1.1 (0.9–1.3)	0.33	1.2 (1.0–1.4)	0.13
Diabetes	1.3 (1.0–1.6)	0.035	1.2 (0.9–1.6)	0.30
CKD	1.1 (0.9–1.4)	0.35	1.0 (0.7–1.3)	0.81

^ap-value for trend, from separate analysis using the CAPA_CYS / LMR_CR ratio as a continuous variable rather than as categorical.

Sensitivity analysis using six eGFR_CYS/eGFR_CR-ratio intervals

When an analysis was undertaken using six, rather than four, eGFR_CYS/eGFR_CR-ratio intervals, it was found that there is a progressive increase of all-cause mortality with a decrease of the ratios starting at the interval 0.70–0.84 (Table 4, Figure 3).

Figure 3. Survival stratified on 6 groups for the ratio of estimated GFR using the CAPA_CYS and LMR_CR equations (six groups; <0.60, 0.60–0.69, 0.70–0.84, 0.85–0.99, 1.00–1.09 and ≥1.10) for all 2781 patients.

Table 4. Cox regression for all-cause mortality in the Lund Cystatin C standardization cohort (n = 2781) using 6 groups for the CAPA_CYS/LMR_CR ratio.

Independent variable	HR (95 % CI)	p	Independent variable	HR (95 % CI)	p
CAPACYS / LMRCR ratio		<0.001^a	mGFR, mL/min/1.73 m^2		0.37
<0.60	3.3 (2.5–4.5)		<30	1.0 (0.6–1.5)
0.60–0.69	2.7 (2.0–3.5)		30–59	0.8 (0.6–1.2)
0.70–0.84	2.2 (1.7–2.8)		60–89	0.9 (0.7–1.2)
0.85–0.99	1.3 (1.0–1.6)		≥90	1.0 (Ref.)
1.00–1.09	1.0 (0.7–1.3)		BMI, kg/m^2		<0.001
≥1.10	1.0 (Ref.)		<20	1.2 (0.9–1.5)
CAPACYS, mL/min/1.73 m^2		0.005	20–24	1.0 (Ref.)
<30	1.6 (0.9–2.7)		25–29	0.8 (0.7–1.0)
30–59	1.0 (0.7–1.6)		≥30	0.7 (0.6–0.8)
60–89	1.1 (0.7–1.6)
≥90	1.0 (Ref.)		Age, per 10 years	1.58 (1.49–1.68)	<0.001
LMRCR, mL/min/1.73 m^2		<0.001	Female	1.1 (1.0–1.3)	0.042
<30	0.7 (0.5–1.2)		Cancer	2.5 (2.2–2.9)	<0.001
30–59	0.5 (0.4–0.8)		CVD	1.1 (1.0–1.3)	0.080
60–89	0.7 (0.5–0.9)		Diabetes	1.2 (1.0–1.5)	0.019
≥90	1.0 (Ref.)		CKD	1.1 (0.9–1.3)	0.52

^ap-value for trend, from separate analysis using the CAPA_CYS / LMR_CR ratio as a continuous variable rather than as categorical.

All-cause mortality using two additional pairs of cystatin C- and creatinine-based GFR estimating equations

Similar results for all-cause mortality of the total cohort were obtained when the eGFR_CYS/eGFR_CR-ratio intervals used to categorize the population were based upon the CKD-EPI_CYS and CKD-EPI_CR equations [26] (Table 5) or the FAS_CYS and FAS_CR equations [27] (Table 6), as when the eGFR_CYS/eGFR_CR-ratio intervals used were based upon the CAPA_CYS-and LMR_CR-equations [19,25].

Table 5. Cox regression for all-cause mortality in the Lund Cystatin C standardization cohort (n = 2781) using eGFR_CYS/eGFR_CR-ratio intervals based on the CKD-EPI equations.

Independent variable	HR (95 % CI)	p	Independent variable	HR (95 % CI)	p
CKD-EPICYS/ CKD-EPICR ratio		<0.001^a	mGFR, mL/min/1.73 m^2		0.174
< 0.70	2.4 (1.9–3.1)		<30	1.0 (0.7–1.6)
0.70–0.84	1.5 (1.2–1.9)		30–59	0.8 (0.6–1.2)
0.85–0.99	1.2 (0.9–1.6)		60–89	0.9 (0.7–1.2)
≥ 1.00	1.0 (Ref.)		≥90	1.0 (Ref.)
CKD-EPICYS, mL/min/1.73 m^2		<0.001	BMI, kg/m^2		<0.001
<30	1.5 (1.0–2.4)		<20	1.3 (1.0–1.6)
30–59	2.5 (1.5–4.1)		20–24	1.0 (Ref.)
60–89	1.4 (1.0–2.0)		25–29	0.8 (0.7–1.0)
≥90	1.0 (Ref.)		≥30	0.7 (0.6–0.8)
CKD-EPICR, mL/min/1.73 m^2		<0.001
<30	0.5 (0.4–0.7)		Age, per 10 years	1.50 (1.41–1.59)	<0.001
30–59	0.7 (0.4–1.0)		Female	1.0 (0.9–1.2)	0.67
60–89	0.6 (0.5–0.7)		Cancer	2.5 (2.2–2.8)	<0.001
≥90	1.0 (Ref.)		CVD	1.1 (1.0–1.3)	0.046
			Diabetes	1.3 (1.1–1.5)	0.007
			CKD	1.1 (0.9–1.3)	0.43

^ap-value for trend, from separate analysis using the CKD-EPI_CYS / CKD-EPI_CR ratio as a continuous variable rather than as categorical.

Table 6. Cox regression for all-cause mortality in the Lund Cystatin C standardization cohort (n = 2781) using eGFR_CYS/eGFR_CRc ratio intervals based on the FAS equations.

Independent variable	HR (95 % CI)	p	Independent variable	HR (95 % CI)	p
FASCYS/ FASCR ratio		<0.001^a	mGFR, mL/min/1.73 m^2		0.36
< 0.70	2.7 (2.2–3.5)		<30	1.3 (0.8–1.9)
0.70–0.84	1.9 (1.5–2.3)		30–59	1.0 (0.7–1.5)
0.85–0.99	1.2 (1.0–1.5)		60–89	1.0 (0.8–1.3)
≥ 1.00	1.0 (Ref.)		≥90	1.0 (Ref.)
FASCYS, mL/min/1.73 m^2		<0.001	BMI, kg/m^2		<0.001
<30	2.2 (1.3–3.7)		<20	1.2 (0.9–1.5)
30–59	1.3 (0.8–2.1)		20–24	1.0 (Ref.)
60–89	1.2 (0.8–1.7)		25–29	0.8 (0.7–0.9)
≥90	1.0 (Ref.)		≥30	0.7 (0.6–0.8)
FASCR, mL/min/1.73 m^2		<0.001
<30	0.9 (0.6–1.4)		Age, per 10 years	1.49 (1.40–1.58)	<0.001
30–59	0.6 (0.4–0.9)		Female	1.1 (1.0–1.3)	0.036
60–89	0.7 (0.6–0.9)		Cancer	2.5 (2.1–2.8)	<0.001
≥90	1.0 (Ref.)		CVD	1.2 (1.0–1.3)	0.034
			Diabetes	1.3 (1.1–1.6)	0.003
			CKD	1.1 (0.9–1.3)	0.27

^ap-value for trend, from separate analysis using the FAS_CYS / FAS_CR ratio as a continuous variable rather than as categorical.

Cause-specific mortality in the total cohort

Cause-specific mortality was studied for the total cohort (Table 7). A decreased eGFR_CYS/eGFR_CR-ratio was associated with higher mortality due all specific causes investigated, but with larger statistical uncertainty in the less common causes of death (DM or CKD).

Table 7. Cox regression for cause-specific mortality in the Lund Cystatin C standardization cohort (n = 2781).

	Cancer mortality		CVD mortality^b		Diabetes mortality^b		CKD mortality^b
	HR (95 % CI)	p	HR (95 % CI)	p	HR (95 % CI)	p	HR (95 % CI)	p
CAPACYS / LMRCR ratio		<0.001^a		0.001^a		0.087^a		0.10^a
< 0.70	3.3 (2.3–4.9)		2.4 (1.6–3.7)		2.0 (0.8–4.7)		5.1 (0.6–46.0)
0.70–0.84	2.4 (1.7–3.3)		1.7 (1.1–2.5)		1.9 (0.8–4.7)		3.3 (0.3–33.7)
0.85–0.99	1.3 (1.0–1.8)		1.2 (0.7–1.8)		1.2 (0.5–3.3)		2.2 (0.2–25.6)
≥ 1.00	1.0 (Ref.)		1.0 (Ref.)		1.0 (Ref.)		1.0 (Ref.)
CAPACYS, mL/min/1.73 m^2		0.21
<30	1.3 (0.6–2.8)		–		–		–
30–59	1.0 (0.6–1.7)		–		–		–
60–89	1.2 (0.8–1.9)		–		–		–
≥90	1.0 (Ref.)		–		–		–
LMRCR, mL/min/1.73 m^2		0.058
<30	0.7 (0.3–1.4)		–		–		–
30–59	0.6 (0.3–0.9)		–		–		–
60–89	0.7 (0.5–0.9)		–		–		–
≥90	1.0 (Ref.)		–		–		–
mGFR, mL/min/1.73 m^2		0.030
<30	0.4 (0.2–0.8)		–		–		–
30–59	0.8 (0.5–1.3)		–		–		–
60–89	0.9 (0.6–1.2)		–		–		–
≥90	1.0 (Ref.)		–
BMI, kg/m^2		0.001		0.80		0.57		0.48
<20	1.1 (0.8–1.5)		0.8 (0.4–1.7)		1.1 (0.1–9.4)		4.7 (0.4–51.8)
20–24	1.0 (Ref.)		1.0 (Ref.)		1.0 (Ref.)		1.0 (Ref.)
25–29	0.8 (0.7–1.0)		0.9 (0.6–1.2)		1.9 (0.8–4.4)		0.8 (0.2–3.9)
≥30	0.7 (0.5–0.9)		0.9 (0.6–1.2)		1.5 (0.6–3.8)		0.7 (0.1–3.7)
Age, per 10 years	1.42 (1.30–1.55)	<0.001	2.54 (2.21–2.94)	<0.001	2.12 (1.58–2.84)	<0.001	3.78 (1.86–7.71)	<0.001
Female	1.2 (1.0–1.5)	0.042	0.7 (0.5–0.9)	0.007	0.8 (0.4–1.4)	0.42	0.5 (0.1–2.1)	0.35
Cancer	7.8 (6.1–9.8)	<0.001	0.7 (0.4–1.0)	0.035	0.4 (0.1–1.2)	0.10	–^c
CVD	1.1 (0.9–1.3)	0.51	1.3 (1.0–1.7)	0.080	1.1 (0.6–2.0)	0.81	4.2 (1.1–16.7)	0.038
Diabetes	0.9 (0.7–1.3)	0.75	1.3 (0.9–1.9)	0.18	8.1 (4.4–15.1)	<0.001	6.7 (1.7–25.5)	0.006
CKD	0.7 (0.5–1.2)	0.26	1.5 (1.1–2.0)	0.011	2.2 (1.2–4.1)	0.011	1.4 (0.4–5.0)	0.64

^ap-value for trend, from separate analysis using the CAPA_CYS / LMR_CR ratio as a continuous variable rather than as categorical.

^bEstimated without CAPA_CYS, LMR_CR and mGFR due to no deaths in the reference category and few cases in the lower categories. The results for the CAPA_CYS /LMR_CR -ratio are similar to those using an unadjusted model.

^cNot able to estimate since no deaths in CKD had a cancer diagnosis.

All-cause mortality for a sub-cohort of persons without diagnoses of cancer, CVD, DM or CKD

All-cause mortality was investigated in a sub-cohort of 1049 persons without any of the disorders studied (cancer, CVD, DM, CKD; overall mortality 21%) at baseline and using the CAPA_CYS- and the LMR_CR-equations. The all-cause mortality in this sub-cohort also increased markedly and continuously with a reduction of the eGFR_CYS/eGFR_CR-ratio and showed an adjusted HR of 3.7 (95% CI 2.2–6.1) when the ratio was <0.70 (Tables 8 and 9).

Table 8. Cox regression for all-cause mortality in three sub-cohorts, patients with no prior diagnosis (n = 1049), patients with normal measured GFR (≥60 mL/min/1.73m²; n = 1300), and patients with both normal mGFR and no prior diagnosis (n = 567).

	Sub-cohort 1		Sub-cohort 2		Sub-cohort 3
	No prior diagnosis		Normal measured GFR		No prior diagnosis and normal measured GFR
	HR (95 % CI)	p	HR (95 % CI)	p	HR (95 % CI)	p
CAPACYS / LMRCR ratio		<0.001^a		<0.001^a		0.001^a
< 0.70	3.7 (2.2–6.1)		4.1 (2.6–6.5)		7.3 (2.3–23.2)
0.70–0.84	2.5 (1.6–4.0)		2.7 (1.9–4.0)		3.2 (1.3–8.0)
0.85–0.99	1.5 (1.0–2.3)		1.6 (1.2–2.3)		1.8 (0.8–4.0)
≥ 1.00	1.0 (Ref.)		1.0 (Ref.)		1.0 (Ref.)
CAPACYS, mL/min/1.73 m^2		0.85		0.69		0.50
<30	0.9 (0.3–3.2)		1.1 (0.4–3.2)		0.4 (0.0–4.4)
30–59	0.8 (0.3–2.3)		0.8 (0.5–1.4)		0.3 (0.1–1.4)
60–89	0.8 (0.3–2.0)		1.0 (0.6–1.5)		0.5 (0.2–1.5)
≥ 90	1.0 (Ref.)		1.0 (Ref.)		1.0 (Ref.)
LMRCR, mL/min/1.73 m^2		0.71		0.063		0.50
<30	1.2 (0.4–3.7)		–		–
30–59	1.0 (0.4–2.5)		0.9 (0.5–1.5)		1.8 (0.4–7.1)
60–89	0.8 (0.4–1.7)		0.7 (0.5–1.0)		1.0 (0.4–2.2)
≥ 90	1.0 (Ref.)		1.0 (Ref.)		1.0 (Ref.)
mGFR, mL/min/1.73 m^2		0.15		0.99		0.19
<30	1.9 (0.7–5.1)		–		–
30–59	1.1 (0.5–2.9)		–		–
60–89	1.5 (0.7–3.3)		1.0 (0.7–1.3)		1.7 (0.8–4.0)
≥ 90	1.0 (Ref.)		1.0 (Ref.)		1.0 (Ref.)
BMI, kg/m^2		0.032		0.019		0.17
<20	1.9 (1.1–3.2)		1.4 (1.0–2.0)		2.5 (1.0–6.0)
20–24	1.0 (Ref.)		1.0 (Ref.)		1.0 (Ref.)
25–29	1.0 (0.7–1.3)		0.9 (0.7–1.1)		0.9 (0.5–1.6)
≥30	0.8 (0.6–1.2)		0.8 (0.6–1.0)		1.2 (0.6–2.5)
Age, per 10 years	1.82 (1.61–2.05)	<0.001	1.45 (1.32–1.60)	<0.001	1.78 (1.42–2.23)	<0.001
Female	1.2 (0.9–1.5)	0.27	1.1 (0.8–1.3)	0.44	1.1 (0.6–1.8)	0.80
Cancer	–		4.3 (3.4–5.5)	<0.001	–
CVD	–		1.0 (0.8–1.3)	0.98	–
Diabetes	–		0.9 (0.5–1.4)	0.58	–
CKD	–		0.6 (0.2–1.8)	0.33	–

^ap-value for trend, from separate analysis using the CAPA_CYS / LMR_CR ratio as a continuous variable rather than as categorical.

Table 9. Total all-cause mortality (%), prevalence (%) of SPS and hazard ratios (HR) for all-cause mortality using different eGFR_CYS/eGFR_CR-ratios for the SPS diagnosis.

	Total cohort		Sub-cohort 1 No prior diagnosis		Sub-cohort 2 Normal measured GFR		Sub-cohort 3 No prior diagnosis and normal measured GFR
CAPACYS / LMRCR ratio	%	HR (95% CI)	%	HR (95% CI)	%	HR (95% CI)	%	HR (95% CI)
<0.70	23.1	3.0 (2.4–3.7)^a	17.1	3.7 (2.2–6.1)^b	17.0	4.1 (2.6–6.5)^b	11.5	7.3 (2.3–23.2)^b
<0.60	11.1	3.3 (2.5–4.5)^c	8.2	4.4 (2.3–8.5)^c	7.4	5.3 (2.8–10.2)^c	5.3	14.2 (2.4–85.2)^c
Total all-cause mortality, %	37.7		21.0		30.2		11.8

Evaluation in the total cohort and in three sub-cohorts, patients with no prior diagnosis (n = 1049), patients with normal measured GFR (≥60 mL/min/1.73m²; n = 1300), and patients with both normal mGFR and no prior diagnosis (n = 567).

^aFully-adjusted model presented in Table 2 (column C).

^bFully-adjusted model presented in Table 8.

^cAdjusted for the same variables as in a and b, respectively, full model not presented.

All-cause mortality for a sub-cohort of persons with normal mGFR regardless diagnosis

It has previously been repeatedly described that all-cause mortality for patients with SPS is also significantly increased in individuals with normal GFR (≥60 mL/min/1.73m²) at baseline [13–17]. But in these studies, only estimated GFR was available. In the present investigation measured GFR was available for all patients, so we used Cox analysis to determine all-cause mortality for the sub-cohort of 1300 patients with normal mGFR regardless diagnosis (≥60 mL/min/1.73m²; overall mortality 30.2%), at baseline. Also, in this sub-cohort the all-cause mortality increased markedly and continuously with a reduction of the eGFR_CYS/eGFR_CR-ratio and showed a HR of 4.1 (95% CI 2.6–6.5), when the ratio was <0.70 (Tables 8 and 9). In this sub-cohort cancer was also associated with an increase in all-cause mortality, but not CVD, DM or CKD (Table 9).

All-cause mortality for a sub-cohort of persons with normal mGFR and no diagnoses of cancer, CVD, DM or CKD

All-cause mortality was investigated in a third sub-cohort of 567 persons with normal mGFR and without any of the disorders studied (overall mortality 11.8%) at baseline. The prevalence of SPS was lower (11.5%) in this cohort than in all other investigated sub-cohorts (Table 9). Nevertheless, the all-cause mortality in this sub-cohort also was markedly and progressively increased with a reduction of the eGFR_CYS/eGFR_CR-ratio and showed an adjusted HR of 7.3 (95% CI 2.3–23) when the ratio was <0.70 (Tables 8 and 9).

Prevalence of SPS in the four cohorts

In most previous studies eGFR_CYS/eGFR_CR-ratios of 0.60 or 0.70 have been used for identifying SPS. The prevalence of SPS, using both ratios, is shown in Table 9 for the studied cohorts, i.e. the total cohort and the three sub-cohorts. The corresponding HRs for all-cause mortality are also shown in Table 9. The prevalence of SPS at baseline was lowest in the sub-cohort of persons with normal mGFR and without any of the investigated diagnoses for both eGFR_CYS/eGFR_CR-ratios used to define SPS, but the corresponding HR for all-cause mortality was the highest.

Influence of eGFR_CYS, eGFR_CR or mGFR on all-cause mortality as a function of the eGFR_CYS/eGFR_CR-ratio

In order to assess the influence of CAPA_CYS, LMR_CR, or mGFR on the eGFR_CYS/eGFR_CR-ratio, these variables were excluded from the regression models, either one at a time or simultaneously. The exclusion of CAPA_CYS, LMR_CR, or mGFR only marginally influenced the HR for all cause-mortality of low ratios of eGFR_CYS/eGFR_CR (Table 10).

Table 10. Cox regression for all-cause mortality in the Lund Cystatin C standardization cohort (n = 2781), when excluding CAPA_CYS, LMR_CR and mGFR, both one at a time and simultaneously.

	p	p	p	p	p	p	p	p
	HR (95 % CI)	HR (95 % CI)	HR (95 % CI)	HR (95 % CI)	HR (95 % CI)	HR (95 % CI)	HR (95 % CI)	HR (95 % CI)
	A	B	C	D	E	F	G	H
CAPACYS / LMRCR ratio	<0.001^a	<0.001^a	<0.001^a	<0.001^a	<0.001^a	<0.001^a	<0.001^a	<0.001^a
< 0.70	3.0 (2.4–3.7)	3.3 (2.7–4.0)	3.0 (2.5–3.7)	3.0 (2.4–3.8)	3.2 (2.6–3.8)	3.3 (2.7–4.0)	3.0 (2.5–3.7)	2.9 (2.4–3.5)
0.70–0.84	2.2 (1.8–2.7)	2.3 (1.9–2.8)	2.3 (1.9–2.8)	2.2 (1.8–2.7)	2.3 (1.9–2.7)	2.3 (1.9–2.8)	2.3 (1.9–2.8)	2.1 (1.7–2.5)
0.85–0.99	1.3 (1.1–1.6)	1.4 (1.1–1.7)	1.3 (1.1–1.6)	1.3 (1.1–1.6)	1.3 (1.1–1.6)	1.4 (1.1–1.7)	1.3 (1.1–1.6)	1.3 (1.1–1.6)
≥ 1.00	1.0 (Ref.)	1.0 (Ref.)	1.0 (Ref.)	1.0 (Ref.)	1.0 (Ref.)	1.0 (Ref.)	1.0 (Ref.)	1.0 (Ref.)

A: adjusted for CAPA_CYS, LMR_CR, mGFR, BMI, age, female, cancer, CVD, diabetes and CKD.

B: same as A but excluding CAPA_CYS.

C: same as A but excluding LMR_CR.

D: same as A but excluding mGFR.

E: same as A but excluding CAPA_CYS and LMR_CR.

F: same as A but excluding CAPA_CYS and mGFR.

G: same as A but excluding mGFR and LMR_CR.

H: same as A but excluding CAPA_CYS, LMR_CR and mGFR.

^ap-value for trend, from separate analysis using the CAPA_CYS / LMR_CR ratio as a continuous variable rather than as categorical.

Discussion

Several studies have indicated that the mean of a cystatin C- and creatinine-based estimation of GFR is more reliable than estimations based upon only one of these parameters [24,28–33]. Estimation strategies based upon this knowledge are therefore used in several laboratories and cater for convenient diagnosis of SPS [33,34]. Elevated concentrations of creatinine or cystatin C have been shown to be related to increased mortality in several populations, but increased cystatin C concentrations have been associated with larger increases in mortality than elevations in creatinine [35–38]. Corresponding observations have been made for decreases of eGFR_CYS or eGFR_CR [36–38]. Although cystatin C might be superior to creatinine as a GFR marker [23,39,40], it has not been possible to show that this is the reason why cystatin C is a better predictor of mortality than creatinine [41]. The presence of a recently proposed kidney disorder, SPS, has been suggested to explain the superiority of cystatin C as a predictor of mortality [11,13–17]. The basic feature of this syndrome is that the renal clearance of 5–40 kDa molecules is decreased more than that of low molecular mass molecules, like water, 0.018 kDa, or creatinine, 0.113 kDa [11,13–18,42]. Previous invasive studies of Beattie and Corcoran [4,5], Roberts et al. [6], Oberbauer et al. [7], Rippe et al. [8] and the pore model for the glomerular filtration barrier [12] offer a pathophysiological interpretation of the basic feature of SPS, namely a decrease in the diameter of a fraction of the pores of the glomerular membrane impairing the filtration of big molecules more than that of small molecules like water, the dominating constituent of the glomerular filtrate, and creatinine. This might also explain why a significant proportion of patients with SPS displays normal GFR, since the molecular masses of the substances used to measure GFR, e.g. ⁵¹Cr-EDTA, iohexol, iothalamate, are much lower (0.344, 0.821, 0.809 kDa) than that of cystatin C (13.3 kDa) used in diagnosing SPS. SPS is characterised by a high cystatin C/creatinine concentration ratio [11,13–18], but more than 20 additional 10–30 kDa proteins display corresponding high protein/creatinine ratios in SPS [11,15,17]. These proteins generally have different, mostly unrelated, functions and their genes are widely distributed among the chromosomes and carry different gene promoters [11,15,17]. It is therefore highly unlikely that their plasma levels are co-regulated by non-renal mechanisms [11] as has been suggested to explain the GFR-independent prognostic value of cystatin C and beta trace protein [43]. A reduction in the common way of excretion of these proteins by the kidneys [1], offers a simple explanation for their parallel increase in concentration characteristic for SPS [11,15,17]. As a matter of fact, in an ongoing investigation of the impact of measured GFR on the human plasma proteome, 2893 proteins are studied in 389 patients with known GFR [44], and preliminary data suggest that many more than 20 proteins are characteristically increased in SPS in parallel with cystatin C.

Although several studies of SPS have shown that it is of clinical relevance, the specific clinical impact of SPS is somewhat unclear, since all studies of it so far have relied on estimated GFR and neither described the diagnoses at baseline, nor the death causes [11,13–18]. The present study is based on a large population in which measured GFR was available for all individuals at baseline. The diagnoses at recruitment were also registered and diagnoses of cancer and CVD dominated, but DM and CKD were also frequent in the cohort, which also comprised individuals considered healthy, e.g. potential kidney donors.

Ratios between eGFR_CYS and eGFR_CR below 0.60 or 0.70 have most frequently been used for diagnosing SPS [11,13–15,18,34]. In the present study SPS was defined by an eGFR_CYS/eGFR_CR-ratio <0.70 as it has been shown produce a high specificity and yet a good sensitivity of a prediction of the 3-year mortality in common types of populations [14]. When this ratio was used to diagnose SPS, adjusted HR for all-cause mortality associated with SPS in the total cohort was higher than the adjusted HRs for all-cause mortality associated with any of the other diagnoses. The HRs associated with SPS were also markedly raised for all the four cause-specific mortality groups of cancer, CVD, DM, or CKD. For CVD mortality, the HR associated with SPS was even higher than the HR of CVD diagnoses. It has previously been shown in studies relying on estimated GFR that a decrease in the eGFR_CYS/eGFR_CR-ratio used to define SPS is associated with an increase in total mortality [13–16] and the same observation is made in the present study adjusting for measured GFR. It might be observed that the most common non-renal influence on the eGFR_CYS/eGFR_CR-ratio, a low muscle mass, does not seem to occur in the present study as BMI does not differ between the groups.

It has previously been shown that SPS is associated with an increase in mortality even among patients with normal GFR (≥60 mL/min/1.73m²), but in these studies only estimated GFR was available [13–16]. The HR for all-cause mortality associated with SPS was markedly increased also in the present study when a sub-cohort of 1300 individuals with normal mGFR was analysed, supporting the idea that the pathophysiological processes of SPS and reduced GFR without SPS differ. The observation that HR for all-cause mortality associated with SPS was high also in a sub-cohort of 1049 patients without diagnoses of CVD, cancer, DM or CKD also indicates that SPS represents a distinct pathophysiological process. The notion that SPS represents a new pathophysiological process is further supported by its high HR for all-cause mortality in the sub-cohort of 567 individuals with normal mGFR and without diagnoses of CVD, cancer, DM or CKD.

In most previous studies of SPS, eGFR_CYS/eGFR_CR-ratios <0.60 or <0.70 have been used to diagnose the syndrome [11,13–15]. In the present study, with known mGFR of the patients, the prevalence of SPS in the total cohort was 23%, when an eGFR_CYS/eGFR_CR-ratio <0.70 was used for diagnosing SPS. Corresponding prevalence in the sub-cohort without diagnoses of CVD, cancer, DM or CKD was 17%, in the sub-cohort with normal mGFR 17%, and in the sub-cohort with normal mGFR and no diagnoses of CVD, cancer, DM or CKD, 12%, respectively. The corresponding HRs for all-cause mortality associated with SPS were all raised and varied between 3.0 and 7.3. It was somewhat unexpected that the highest HR was observed for the sub-cohort with normal mGFR and without any diagnosis. This sub-cohort has a total all-cause mortality of 11.8%, which is much lower than that of the total cohort (37.7%) and represents the healthiest of the investigated cohorts. The high prevalence of SPS in this sub-cohort also indicates that SPS represents a unique pathophysiological process. The absence of any of the investigated disorders in the healthiest sub-cohort might augment the relative influence of an SPS diagnosis on the all-cause mortality, producing a high HR.

It has earlier been shown that a decrease in eGFR_CYS is associated with an increase in mortality [36–38]. This is also demonstrated in the present investigation, but the adjusted HR for all-cause mortality of SPS is considerably higher than the corresponding adjusted HR associated with a decrease in eGFR_CYS. This agrees with the present study of the influence of CAPA_CYS, LMR_CR, or mGFR on the adjusted HR for all-cause mortality of SPS, showing that exclusion of any of these variables only marginally influences HR.

To diagnose SPS, two GFR-estimating equations based upon cystatin C or creatinine are required. As previously shown and corroborated in this study, not only the equation pair CAPA_CYS and LMR_CR, but also the equation pairs CKD-EPI_CYS and CKD-EPI_CR, or FAS_CYS and FAS_CR are able to define cohorts with increased mortality and/or morbidity due to SPS [13,14,16–18]. We prefer equation pairs such as CAPA_CYS and LMR_CR, because they work for both children and adults [19,30,34,45], which is important since emerging evidence indicates that SPS is also present in children [46,47].

Diagnosing kidney disorders is presently mainly based upon determination or estimation of GFR, specifying the volume of primary urine produced per unit of time. The dominating part of this is water so the conventional way of diagnosing kidney disorders does not consider the important role of the kidneys in maintaining a dynamic equilibrium of 5–40 kDa molecules in extracellular fluids. Proteins with signalling functions constitute a significant part of these molecules and in SPS the decrease in their clearance will induce a general increase in their concentration in extracellular fluids with a consequent disturbance of the signalling system in the body. Recent studies of the proteomes of patients with and without SPS at normal or reduced mGFR, support this proposal for the pathophysiological mechanism in SPS [15]. This mechanism means that the composition, or quality, of the glomerular filtrate is changed in SPS and it has previously been suggested that a similar mechanism is involved in the alteration of the glomerular filtration process occurring in normal pregnancy and, more pronounced, in preeclampsia [9–11]. Earlier invasive studies have also indicated changes in the composition of the glomerular filtrate in other kidney disorders [4–8]. It is interesting that the suggested pathophysiology of SPS allows for treatment modalities aiming at reducing the high levels of disease-producing signal proteins by use of, for example, monoclonal antibodies [15].

The present study, showing that SPS strongly increases mortality in cohorts with diagnoses of CVD, cancer, DM or CKD, as well as in cohorts with normal measured GFR and without these diagnoses, indicates that a significant part of all individuals with serious kidney disorders might be missed by the conventionally used definition and classification of CKD recommended by KDIGO in 2012 [48]. This might suggest that optimal classification and stratification of CKD requires not only analysis of GFR (estimated or measured) and the urine albumin/creatinine ratio, but also determination of the eGFR_CYS/eGFR_CR-ratio to assess the presence of SPS.

The present study has some limitations. Although the data in the Swedish patient register are generally of high quality [49] some misclassification of disease history is probably present at baseline in our study. The diagnosis of CKD from the registers is based on ordinary clinical work-up where clinicians use the KDIGO 2012 CKD classification for CKD stages 1 to 5. Information about albuminuria is not always included, which means that we most likely to some extent underestimate the prevalence of CKD at baseline. However, we find it unlikely that this misclassification is dependent on SPS status. It should also be noted that the study investigates SPS as a predictor of mortality, and that no firm conclusion about the causality of the observed associations can be drawn based on our results. In particular residual confounding from e.g. other aspects of health not present in the registers cannot be ruled out. We addressed this limitation to some extent by investigating the robustness of the results across sub-cohorts likely to be more similar with respect to unmeasured health determinants. Finally, it should be noted that the present study was based on a selected cohort of patients with diverse medical reasons for having their GFR measured. The findings are thus not necessarily generalizable to general or healthier populations.

Acknowledgments

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

Disclosure statement

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