Association of GSTM1 Deletion With Progression of CKD in Children: Findings From the Chronic Kidney Disease in Children (CKiD) Study

Abstract

Rationale & Objective:

Loss of function of the product of the GSTM1 gene has been implicated in rapid progression of adult chronic kidney disease (CKD). Its role in pediatric CKD has not been previously described.

Study Design:

Secondary analysis of a prospective observational cohort examining the association between deletions in GSTM1 and progression of CKD.

Setting & Participants:

We used data and samples from the prospective Chronic Kidney Disease in Children (CKiD) cohort aged 1-16 years at enrollment with CKD.

Exposure:

We defined the exposure as fewer than 2 GSTM1 alleles on real-time polymerase chain reaction amplification.

Outcome:

The primary outcome was a composite of 50% decrease in estimated glomerular filtration rate (eGFR) or start of kidney replacement therapy. Secondary outcomes included remission of proteinuria in children with glomerular disease and cardiovascular complications.

Analytical Approach:

The primary analysis was by Cox proportional hazards model. Analysis was adjusted for age, sex, race, ethnicity, body mass index category, diagnosis category, and eGFR.

Results:

The analysis included 674 children. Their mean age at most recent visit was 11.9 years; 61% were male, and 20% were Black. There were 241 occurrences of the primary outcome at the time of analysis. After adjustment for baseline characteristics, the risk of progression of CKD for exposed children was 1.94 (95% CI, 1.27-2.97). The effect size was similar with either 1 or 2 deletions (autosomal dominant inheritance). The relationships between number of functional GSTM1 alleles and prespecified secondary outcomes were not statistically significant after adjustment.

Limitations:

Missing data, especially for secondary outcomes, and relatively small sample size compared to genetic studies in adults.

Conclusions:

GSTM1 deletion is associated with more rapid progression of pediatric CKD after adjustment in this large prospective cohort. No statistically significant associations were seen with secondary outcomes. If replicated, these findings may inform development of interventions for CKD in children.

Chronic kidney disease (CKD) leads to significant morbidity and mortality in children.¹ Progression of CKD to incident kidney failure with replacement therapy (KFRT) is mediated by several factors, including patient characteristics (male sex, APOL1 high-risk gene variants)² and disease characteristics (glomerular disease, proteinuria, abnormal blood pressure).³ Sequelae of pediatric CKD include poor quality of life and cardiovascular disease.

Increased oxidative stress is linked to CKD progression in adults.⁴ Recent research has implicated the glutathione S-transferase family of genes, specifically GSTM1, in the regulation of these oxidative stress pathways. In the general population, 53% of White Europeans and North Americans and 27% of Africans are homozygous for the null allele of this gene.⁵ In adults, absence of GSTM1 is associated with an increased risk of susceptibility to KFRT compared with healthy controls in multiple populations, including Mexicans⁶ and North Indians.⁷ In African Americans with CKD attributable to hypertension, absence of GSTM1 is associated with faster progression to KFRT.⁸ Of note, this study found that GSTM1 status modified the effect of APOL1 on progression, increasing the odds among those with the APOL1 high-risk variant from 2.13 (95% CI, 0.76-5.90) for those with the GSTM1 functional variant to 3.0 (95% CI, 1.51-5.96) for those with the GSTM1 null allele.⁹

In adults, there is also a significant link between absence of GSTM1 product and incidence of coronary artery disease^10,11 as well as heart failure.¹² In pediatrics, less is known about the effect of GSTM1 null allele, but GSTM1 deficiency appears to play a role in childhood asthma¹³ and development of acute lymphoblastic leukemia.¹⁴ To date, there is no information on the role of this important system in CKD in children. Thus, we undertook this study in the well-characterized Chronic Kidney Disease in Children (CKiD) cohort to evaluate the role of GSTM1 gene variants in kidney disease progression and sequelae. We hypothesized that participants with at least 1 nonfunctional GSTM1 allele would have faster progression of CKD than participants with 2 functional alleles.

Methods

Study Population

The CKiD study is a multicenter, prospective, observational cohort whose purpose is to determine the risk factors for decline in kidney function and to define the cardiovascular, metabolic, and neurocognitive effects of progressive CKD in children.¹⁵ Enrollment began in 2005 and is ongoing. Children aged 1-16 years are eligible for enrollment in the study if they have an estimated glomerular filtration rate (eGFR) between 30-90 mL/min/1.73 m². CKiD exclusion criteria include organ transplantation, cancer or HIV virus infection, current or recent pregnancy, genetic syndromes, current enrollment in a randomized clinical trial, history of structural heart disease, or inability to complete data collection or follow-up evaluations. All data including demographics were anchored to visit 2, the first follow-up visit after enrollment, at which time baseline testing was performed. Participants were censored on date of death, at last visit if lost to follow-up after 1.75 years, or if event-free at the time of analysis (after January 15, 2020).

Exposure

GSTM1 status was estimated using a TaqMan Copy Number Assay (Hs02575461_cn; ThermoFisher).¹⁶ Briefly, real-time polymerase chain reaction was performed in a 5 μL reaction under conditions described by the manufacturer using ribonuclease P as an internal control. Samples were run in quadruplicate on an Applied Biosystems 7900-HT real-time instrument and scored as 0, 1, or 2 alleles using Applied Biosystems Copycaller software (version 2.1). Only samples that worked 3 or more times and were always concordant were scored.

The exposure in this analysis was deletion in the GSTM1 gene leading to loss of protein production. The ability to produce functional GSTM1 protein was considered the effect allele; therefore, the haplotypes are referred to as 0/0 (2 alleles with deletion), 0/1 (heterozygous), and 1/1 (2 undeleted alleles). This was a priori specified as a dichotomous exposure: 2 functional alleles versus at least 1 null allele, consistent with the current literature.⁹

Outcomes

The prespecified primary outcome of this analysis is an outcome of the main CKiD study: a combination of 50% decrease in eGFR or KFRT, defined as start of dialysis or transplantation. We estimated eGFR using equations validated for use in reduced kidney function, using creatinine and cystatin C measured at every study visit (annually).^15,17 In the subset of participants with glomerular disease, remission of proteinuria was analyzed as a secondary outcome. In these participants, remission of proteinuria was defined as the combination of complete (urinary protein-creatinine ratio [UPCR] < 0.2 mg/mg) or partial (50% decrease in UPCR).

For all participants, cardiovascular secondary outcomes were assessed. Blood pressure was measured 3 times in 1 sitting at a study clinic visit¹⁸; hypertension was defined according to 2017 American Academy of Pediatric Clinical Practice Guidelines.¹⁹ Left ventricular mass was measured by transthoracic echocardiography read at the central Cardiovascular Core Imaging Research Laboratory (CCIRL, Cincinnati, OH) and indexed to body size; left ventricular hypertrophy (LVH) was defined as left ventricular mass index > 95th percentile for age.²⁰ Carotid intima-medial thickness (cIMT) was likewise imaged at individual clinical sites and read at the CCIRL.²¹

Covariates

Race was self-described in a written questionnaire. The options were prespecified by the CKiD investigators and categorized for this analysis as non-Black or Black. Similarly, ethnicity was self-described and categorized in this analysis as non-Hispanic or Hispanic. Body mass index (BMI) was normalized as a z score and categorized as normal (<85th percentile), overweight (85th-<95th percentile), or obese (≥95th percentile).

Primary kidney disease diagnosis was classified as glomerular versus nonglomerular, adjudicated by the CKiD Central Steering Committee; glomerular diseases included hemolytic-uremic syndrome, focal segmental glomerulosclerosis, familial nephritis, IgA nephropathy, systemic immunological disease, chronic glomerulonephritis, membranoproliferative glomerulonephritis, idiopathic crescentic glomerulonephritis, membranous nephropathy, Henoch-Schönlein nephritis, congenital nephrotic syndrome, sickle cell nephropathy, diabetic nephropathy, and other nonspecified glomerular disease (assigned by the enrolling physician).

Baseline kidney function was assessed as estimated glomerular filtration rate (eGFR) using the combined creatinine–cystatin C equation validated in this cohort.¹⁷ Quality of life was assessed using the 23-item PedsQL (Pediatric Quality of Life Inventory), a generic health-related quality of life instrument²² that has been specifically validated in the pediatric CKD population.²³ The PedsQL was administered to all participants aged 8 years and older and to parental proxy for all ages.

We did not adjust for proteinuria in the main analysis nor for other biochemical markers to avoid adjusting for known mediators of progression of CKD. This is consistent with other research using the CKiD cohort.^24,25

Statistical Analysis

Analysis was conducted of complete cases. Continuous variables were described by median and interquartile range (IQR); group differences were tested by Mann-Whitney U test. Categorical variables and Hardy-Weinberg equilibrium were described as percentages and compared by χ² test. Analysis of the association of primary end point, progression of CKD, with the exposure of interest was conducted by Cox proportional hazards model. The models were developed in a forward stepwise fashion. Model 1 was completely unadjusted; model 2 was adjusted for age, sex, and BMI z score; and model 3 was further adjusted for diagnosis category and baseline kidney function.

Adjusted survival curves were plotted by estimating hazard ratios for participants with continuous covariates set to the mean and categorical covariates set to an arbitrary reference category (female, nonglomerular disease). Subgroup analyses by sex and underlying kidney disease diagnoses were conducted similarly. Sensitivity analyses were also conducted with an alternative dichotomous exposure (no functional alleles versus at least 1 functional allele) and haploinsufficient exposure (0 vs 1 vs 2 null alleles) inheritance models, as well as with the inclusion of baseline proteinuria. Examination of the association of GSTM1 deletion with remission of proteinuria was performed similarly by Cox proportional hazards model. The outcomes of LVH and hypertension were examined using multivariable logistic regression. The association of the exposure with cIMT was examined using multivariable linear regression. All associations were adjusted for the covariates described previously.

Ethics

The main CKiD study (ClinicalTrials.gov identifier NCT00327860) has been continuously approved by all individual sites’ institutional review boards (IRB) since inception. This analysis was separately approved by the Albert Einstein College of Medicine (IRB approval #2018-9847). Written informed consent was obtained from parent/guardians of pediatric participants of the main CKiD study, with age-appropriate assent according to local IRB guidance. This report was drafted in accordance with STREGA guidelines for reporting cohort studies.²⁶

Results

Participant Characteristics

Of the 866 participants enrolled in CKiD, a total of 674 children were included in the analysis after exclusion for missing DNA samples, missing covariates, and unsuccessful DNA amplification. Details of how the participants were identified from the main CKiD enrollment are shown in Figure 1. The average age at the first follow-up visit was 11.9 years; 61% were male, and 20% were Black. The characteristics of study participants by GSTM1 deletion status are shown in Table 1. The participants were not statistically significantly different in most demographic or clinical findings at baseline. Participants with GSTM1 deletions were less likely to be Black (18% vs 36%) than the participants with 2 functional alleles. Stratified by race, the population was in Hardy-Weinberg equilibrium (χ² White P = 0.34, Black P = 0.86). There were a total of 241 events of the primary outcome (50% decrease in eGFR or incident KFRT) at the time of analysis; there were 38 instances of the secondary outcome (UPCR < 0.2 mg/mg or 50% reduction in UPCR) in the group with glomerular disease. The median follow-up time was 5.2 (IQR, 2.7-7.3) years.

Figure 1.

Flowchart showing study participant selection. Of 866 participants enrolled in the CKiD cohort between 2005 and 2019, 674 remained after fulfilling inclusion and exclusion criteria. Abbreviations: CKiD, Chronic Kidney Disease in Children; PCR, polymerase chain reaction.

Table 1.

Baseline Characteristics of 674 Participants of CKiD by GSTM1 Status

	Total (N = 674)	GSTM1 0/0 or 0/1 (n = 605)	GSTM1 1/1 (n = 69)	P
Age, y	11.9 [8.0-15.7]	10.7 [6.8-14.9]	11.9 [8.3-15.7]	0.1
Male sex	412 (61%)	367 (65%)	45 (61%)	0.5
Black race	136 (20%)	111 (18%)	25 (36%)	<0.001
Hispanic ethnicity	89 (13%)	78 (13%)	11 (15%)	0.5
BMI category				0.5
Normal	477 (71%)	431 (71%)	46 (67%)
Overweight	110 (16%)	99 (16%)	11 (16%)
Obese	87 (13%)	75 (12%)	12 (17%)
Primary disease				0.9
HUS	33 (4.9%)	29 (4.8%)	4 (5.8%)
Glomerular (non-HUS)	158 (23%)	143(24%)	15 (22%)
Genitourinary/cystic/hereditary	438 (65%)	392 (65%)	46 (67%)
Nonglomerular other	46 (6.7%)	41 (6.8%)	4 (5.8%)
Baseline eGFR (mL/min/1.73 m2)	53 [38-68]	55 [39-68]	46 [36-66]	0.07
UPCR (mg/mg)	0.32 [0.12-0.98]	0.32 [0.12-0.93]	0.34 [0.13-1.07]	0.5
Hypoalbuminuria (n = 664)	82 (12%)	70 (12%)	12 (18%)	0.1
Dyslipidemia (n = 585)	278 (48%)	252 (48%)	26 (46%)	0.8
Anemia (n = 655)	179 (27%)	161 (27%)	18 (27%)	0.9
High uric acid (n = 358)	177 (49%)	163 (50%)	14 (47%)	0.8
PedsQL score (n = 595)	80 [67-91]	82 [68-91]	78 [59-87]	0.08
Parent QoL score (n = 452)	79 [68-89]	79 [68-89]	78 [65-86]	0.3
Left ventricular hypertrophy	358 (53%)	326 (54%)	32 (46%)	0.2
Blood pressure (n = 648)				0.9
Normal	414 (64%)	373 (64%)	41 (61%)
Elevated	82 (12%)	74 (13%)	8 (12%)
Stage 1 hypertension	122 (19%)	110 (19%)	15 (22%)
Stage 2 hypertension	27 (4.2%)	24 (4.1%)	3 (4.5%)
cIMT, mm (n = 140)	0.42 [0.36-0.47]	0.42 [0.37-0.47]	0.35 [0.33-0.49]	0.8

Continuous variables are described as median [interquartile range]; categorical variables are described as n (%). Abbreviations: BMI, body mass index; cIMT, carotid intima-medial thickness; CKiD, Chronic Kidney Disease in Children Study; eGFR, estimated glomerular filtration rate; HUS, hemolytic-uremic syndrome; PedsQL, Pediatric Quality of Life Inventory; QoL, quality of life; UPCR, urinary protein-creatinine ratio.

Association of GSTM1 Deletion with Progression of CKD

The association of GSTM1 deletion with the primary CKiD outcome, either 50% decrease in eGFR or incident KFRT, is shown in Table 2.

Table 2.

Association of GSTM1 Deletion Status With Progression of CKD in 674 Participants of CKiD

	GSTM1 0/0 or 0/1 (n = 605)			GSTM1 1/1 (n = 69)			HR (0/0 or 0/1 vs 1/1)	P
	No. of Events	FU, y	Event Rate, y−1	No. of Events	FU, y	Event Rate, y−1
Model 1a	216	5.2 [2.7-7.2]	0.067 (0.058-0.076)	25	5.4 [2.8-9.5]	0.062 (0.042-0.091)	1.11 (0.73-1.68)	0.6
Model 2b							1.04 (0.69-1.58)	0.9
Model 3c							1.94 (1.26-2.98)	0.002
Sexc								0.9d
Male (n = 412)	129	5.4 [2.9-7.2]	0.065 (0.055-0.078)	17	5.5 [2.8-9.5]	0.064 (0.040-0.100)	2.03 (1.18-3.48)	0.01
Female (n = 262)	87	5.1 [2.3-7.5]	0.069 (0.056-0.085)	8	5.3 [2.7-8.9]	0.058 (0.029-0.120)	1.70 (0.82-3.52)	0.2
Etiologyc								0.9d
Nonglomerular (n = 483)	150	6.1 [3.0-7.7]	0.060 (0.051-0.070)	19	6.4 [3.3-9.7]	0.058 (0.037-0.091)	2.03 (1.23-3.34)	0.005
Glomerular (n = 191)	66	4.1 [1.8-6.0]	0.092 (0.072-0.120)	6	2.8 [1.1-5.4]	0.077 (0.035-0.170)	1.73 (0.73-4.10)	0.2

Follow-up time is described as median [interquartile range]. Event rates and hazard ratios show associated 95% confidence intervals. Abbreviations: BMI, body mass index; CKD, chronic kidney disease; CKiD, Chronic Kidney Disease in Children Study; FU, follow-up; HR, hazard ratio.

^aUnadjusted.

^bAdjusted for age, sex, and BMI z score.

^cAdjusted for age, sex, BMI z score, diagnosis category, and baseline kidney function.

^dP for interaction.

In unadjusted analysis, the association between GSTM1 deletion was not statistically significant, with hazard ratio (HR) of 1.03 (95% CI, 0.69-1.54). However, after adjustment for baseline eGFR, a statistically significant relationship emerged, with a HR of 1.94 (95% CI, 1.26-2.98) (Fig 2). Stratification by race and diagnosis showed statistically significant relationships for participants with nonglomerular disease (HR, 2.03 [95% CI, 1.23-3.34]) (Table 2) but not for those with glomerular diseases.

Figure 2.

Decrease of 50% in eGFR or incident kidney failure with replacement therapy by GSTM1 deletion status in CKiD participants. Curves are for a hypothetical participant with average values of continuous predictors (age, BMI, and baseline kidney function) and reference categories of categorical variables (female sex, nonglomerular disease).¹ Abbreviations: BMI, body mass index; CKiD, Chronic Kidney Disease in Children; LR, likelihood ratio.

Sensitivity analysis defining exposure using different inheritance models showed statistical significance in both the a priori dichotomous model and a haploinsufficient model after adjustment for age, sex, BMI, diagnosis category, and baseline kidney function (Table 3). The effect size with either 1 or 2 nonfunctional alleles was similar, consistent with autosomal dominant inheritance. The main effect of GSTM1 deletion on progression of CKD was not seen after adjustment for baseline proteinuria (Table S1).

Table 3.

Progression of CKD by Varying Inheritance Model of GSTM1 Deletion in CKiD Participants, Adjusted for Age, Sex, Race, Ethnicity, BMI Category, Diagnosis Category, and Baseline Kidney Function

	No. of Events	FU Time, y	Event Rate, y−1	HR	P
Haploinsufficient model					0.05
GSTM1 1/1 (n = 69)	25	5.4 [2.8-9.5]	0.062 (0.042-0.091)	1.00 (reference)
GSTM1 0/0 (n = 306)	108	5.5 [2.8-7.3]	0.064 (0.053-0.077)	1.86 (1.18-2.91)	0.007
GSTM1 1/0 (n = 299)	108	5.0 [2.5-7.2]	0.070 (0.058-0.085)	2.04 (1.30-3.20)	0.002
Recessive model
GSTM1 1/0 or 1/1 (n = 368)	133	5.1 [2.6-7.4]	0.068 (0.058-0.081)	1.00 (reference)
GSTM1 0/0 (n = 306)	108	5.5 [2.8-7.3]	0.064 (0.053-0.077)	1.08 (0.83-1.40)	0.6

Follow-up time is described as median [interquartile range]. Event rates and hazard ratios show associated 95% confidence intervals. Abbreviations: BMI, body mass index; CKD, chronic kidney disease; CKiD, Chronic Kidney Disease in Children Study; FU, follow-up; HR, hazard ratio.

Association between GSTM1 Deletion and Secondary Outcomes

We next tested the association of remission of proteinuria with GSTM1 deletion in study participants who had glomerular disease. In this subpopulation, there was no statistically significant association. There was also no statistically significant association between GSTM1 deletion and LVH, hypertension, or cIMT (Table 4).

Table 4.

Association of GSTM1 Deletion Status With Secondary Outcomes, Adjusted for Age, Sex, Race, Ethnicity, BMI Category, Diagnosis Category, and Baseline Kidney Function in CKiD Participants

	No. of Events	Outcome	P
Remission of proteinuria (n = 182)a	38	1.23 (0.40, 3.81)	0.7
Cardiovascular outcomes
Left ventricular hypertrophy (n = 674)b	358	1.29 (0.70, 2.40)	0.4
Hypertension (n = 648)b	152	0.97 (0.53, 1.77)	0.9
Natural logarithm cIMT (n = 140)c		0.050 (−0.16, 0.26)	0.6

Values in parentheses are 95% confidence interval. Abbreviations: BMI, body mass index; cIMT, carotid intima-medial thickness; CKiD, Chronic Kidney Disease in Children Study.

^aHazard ratio. Additionally adjusted for baseline proteinuria.

^bLogistic regression odds ratio.

^cLinear regression coefficient.

Discussion

In this analysis, we found a statistically significant association between GSTM1 status and progression of CKD in children enrolled in the CKiD study. This association remained significant among non-Black participants and among those with nonglomerular kidney disease. The association was seen with an autosomal dominant inheritance model, without dose effect in haploinsufficiency. There was no statistically significant association between GSTM1 status and remission of proteinuria or cardiovascular measures.

The enzyme product of the GSTM1 gene participates in detoxification of a wide variety of substrates. GSTM1 deletions leading to loss of product are common in multiple populations and have been associated with a variety of adverse health outcomes, including cancer, cardiovascular disease, and asthma. In terms of kidney disease, GSTM1 deletions are associated with both incidence of KFRT and progression of CKD in ethnically diverse adult populations.^6–8 This new analysis shows similar findings in children. This was not necessarily a foregone conclusion because the etiologies of adult and pediatric kidney disease are fundamentally different. Adult-onset CKD in middle- and high-income countries most often results from hypertensive nephrosclerosis or diabetic nephropathy.²⁷ In pediatrics, congenital anomalies of the kidney and urinary tract and hereditary nephropathies are responsible for more than half of CKD cases.²⁸ The similar effects of GSTM1 deletion in adults and children suggest a common pathway of CKD progression involving oxidative damage despite different initial insults.

The effect of GSTM1 deletion was not seen until after adjustment for covariates, specifically enrollment eGFR, which differed between groups. Participants with 2 functional GSTM1 alleles were more likely to be Black and had worse kidney function at enrollment than participants with 1 or 2 nonfunctional alleles. These 2 features are known to be independent risks for progression of kidney disease, and this may have confounded the unadjusted results.

In subgroup analysis, effect size was similar between sexes. We did not find a statistically significant association between GSTM1 deletion status and progression of CKD in participants with glomerular disease. Although adult studies have seen such an association,^9,29 the smaller size of these subgroups in the CKiD cohort limited analytic power. Further studies in populations enriched for these characteristics are necessary to fully describe these associations. Likewise, the lack of effect after adjustment for proteinuria can be attributed to mediation.

Currently, no specific interventions are approved for delaying or reversing progression of pediatric CKD. Identification of children with risk factors for rapid progression would allow for closer monitoring, potentially including more prompt treatment of complications and preparation for kidney replacement therapy. To date, specific treatment of oxidative damage associated with absence of GSTM1 protein has not been tested. Increased consumption of sulforaphane, either through cruciferous vegetable intake or supplementation, has been shown to protect against progression of kidney disease in both mouse models and human adult studies.³⁰ Further studies, including randomized controlled trials are needed to determine whether these types of treatments are safe and effective in children with CKD.

We did not see statistically significant associations of secondary outcomes including remission of proteinuria in glomerular disease or cardiovascular sequelae with GSTM1 deletion. These complications contribute to progression of pediatric CKD³ and, in the case of the cardiovascular complications, mortality.¹ Identification of new markers, including genetic markers, of cardiovascular mortality are needed to identify children with CKD in need of closer monitoring and management. As discussed previously, the subgroup of children with glomerular disease was likely underpowered to examine associations with proteinuria. In animal models, absence of GSTM1 protein is associated with increased vascular smooth muscle cell proliferation and migration leading to renal vascular injury with medial hypertrophy and hyperplasia.³¹ Clinical associations of GSTM1 absence with hypertension and atherosclerosis have been seen in studies of adults with risk factors (eg, diabetes mellitus, smoking) in a 2-hit fashion,^32,33 but no similar data have been published in pediatric populations or kidney disease populations.

As this was a secondary retrospective analysis of the CKiD cohort, we were unable to collect new data of interest or verify data. There were substantial missing data for some laboratory analyses (eg, uric acid) and advanced testing including cIMT. Race was categorized as Black or non-Black, limiting evaluation of other racial backgrounds in which GSTM1 variants may be more or less common. Furthermore, diagnosis category was reported by parents and therefore may be prone to misclassification bias.

Because our cohort was composed of participants with diagnosed CKD, we were unable to estimate the effect of GSTM1 deletion on incidence of CKD. Although CKiD is the largest cohort of children with CKD available for analysis, it remains small compared with studies of adults, especially in analysis of undersampled subgroups including racial minorities. For this reason, we were unable to confidently analyze the GSTM1-APOL1 interaction seen in adult studies. Furthermore, we were limited by the size of the cohort in the number of covariates included in models to prevent overfitting.

The strengths of this study include its well-phenotyped cohort with prospectively collected data and long follow-up times. All laboratory measures were analyzed centrally, and imaging and clinical data were collected according to standardized protocols. GSTM1 deletion status was determined centrally by a laboratory experienced in analyzing this gene; GSTM1 is located in an area of the genome with significant homology, so there is potential for genotyping error.

In conclusion, we report a previously undescribed association between GSTM1 deletion and progression of pediatric CKD in a large prospective cohort. This is consistent with current literature of adults with CKD and animal models. This finding may guide clinicians in monitoring and surveillance of patients with CKD for development of sequelae and prognosis of need for kidney replacement therapy. Further studies are needed to assess the association of this genetic variant with progression of disease in specific patient groups and disease subtypes.

Supplementary Material

1

Table S1: Sensitivity analysis of the association of GSTM1 copy number with CKD progression, adjusted for age, sex, BMI z score, diagnosis category, baseline kidney function, and baseline proteinuria.

PLAIN-LANGUAGE SUMMARY.

Although chronic kidney disease is rare in children, it may cause significant harm when it occurs. In this study, we used a multicenter cohort of children with chronic kidney disease to test the association of GSTM1 deletion with progression of kidney disease. We found that children with nonfunctional GSTM1 alleles were almost twice as likely to lose half their kidney function or to need kidney replacement therapy as were children with 2 functioning alleles. Differences in GSTM1 status were not associated with cardiovascular outcomes or remission of proteinuria. This finding could help to identify children who could benefit from closer monitoring.