Genetic spectrum of CAKUT and risk factors for kidney failure: a paediatric multicentre cohort study

ABSTRACT

Background

Congenital anomalies of the kidney and urinary tract (CAKUT) are the leading cause of kidney failure in children with phenotypic and genotypic heterogeneity. Our objective was to describe the genetic spectrum and identify the risk factors for kidney failure in children with CAKUT.

Methods

Clinical and genetic data were derived from a multicentre network [Chinese Children Genetic Kidney Disease Database (CCGKDD)] and the Chigene database. A total of 925 children with CAKUT who underwent genetic testing from 2014 to 2020 across China were studied. Data for a total of 584 children were obtained from the CCGKDD, including longitudinal data regarding kidney function. The risk factors for kidney failure were determined by the Kaplan–Meier method and Cox proportional hazards models.

Results

A genetic diagnosis was established in 96 of 925 (10.3%) children, including 72 (8%) with monogenic variants, 20 (2%) with copy number variants (CNVs) and 4 (0.4%) with major chromosomal anomalies. Patients with skeletal abnormalities were more likely to have large CNVs or abnormal karyotypes than monogenic variants. Eighty-two patients from the CCGKDD progressed to kidney failure at a median age of 13.0 years (95% confidence interval 12.4–13.6) and 24 were genetically diagnosed with variants of PAX2, TNXB, EYA1, HNF1B and GATA3 or the 48,XXYY karyotype. The multivariate analysis indicated that solitary kidney, posterior urethral valves, bilateral hypodysplasia, the presence of certain variants and premature birth were independent prognostic factors.

Conclusions

The genetic spectrum of CAKUT varies among different subphenotypes. The identified factors indicate areas that require special attention.

Graphical Abstract

Graphical Abstract.

KEY LEARNING POINTS.

What is already known about this subject?

• Congenital anomalies of the kidney and urinary tract (CAKUT) are the leading cause of kidney failure in children.

• Few analyses, including genetic diagnoses and detailed analyses of the clinical phenotypes, have been performed on large cohorts to determine the risk factors for kidney failure.

What this study adds?

• Children present a wide genetic spectrum and complicated genotype–phenotype correlations among different CAKUT subphenotypes.

• A solitary kidney; posterior urethral valves; bilateral hypodysplasia; variants in PAX2, TNXB, EYA1, HNF1B and GATA3 and the 48, XXYY karyotype; and premature birth are associated with an increased risk of kidney failure during childhood.

What impact this may have on practice or policy?

• Patients with skeletal abnormalities may benefit more from a CNV analysis and karyotype testing. The identified factors indicate areas that require special attention in children with CAKUT.

INTRODUCTION

Congenital anomalies of the kidney and urinary tract (CAKUT) occur in 3–6 per 1000 live births [1] and are the most common causes of chronic kidney disease (CKD) and kidney failure (29–40%) in children, as reported in different registries [2, 3]. CAKUT has a multifactorial aetiology involving genetic, epigenetic and environmental factors [4, 5]. This disorder is characterized by high phenotypic variability between patients and the co-occurrence of different anomalies and extrarenal manifestations within the same individual [5, 6]. Monogenic and copy number variants (CNVs) involved in syndromic CAKUT may also lead to isolated CAKUT [7].

The prognosis of CAKUT is mainly related to the extent of nephron number reduction and the associated risk of kidney failure [7]. Several studies have explored risk factors associated with kidney failure [8–12]. One large study showed that a solitary kidney and renal hypodysplasia (RHD) associated with the posterior urethral valve (PUV) are independent risk factors [12]. In univariate analyses of genetic risk factors for CAKUT associated with kidney failure, certain variant carriers experienced worse outcomes [9, 10, 13].

Few multivariate analyses including genetic diagnosis have been performed on large cohorts of children with CAKUT to determine the risk factors for kidney failure [11, 14]. The aim of this study was to describe the genetic spectrum of different CAKUT phenotypes and evaluate risk factors for kidney failure.

MATERIALS AND METHODS

Study design and participants

Patients diagnosed with CAKUT based on kidney imaging studies who were genetically tested before the age of 18 years included in the Chinese Children's Genetic Kidney Disease Database (CCGKDD) and Chigene database between 2014 and 2020 were recruited. Data on demographics, genetic diagnosis, genetic test methods, CAKUT and extrarenal phenotypes were collected for further analysis.

The study consisted of two parts and the outline is illustrated in Supplementary Fig. S1. We evaluated the correlations between clinical presentation patterns (sex, family history, genetic test methods, categories of CAKUT and extrarenal anomalies) and genetic abnormalities in part 1. In part 2, the risk factors associated with kidney failure were evaluated in patients from the CCGKDD, including genotypes, CAKUT phenotypes and birth history.

The CCGKDD is a national multicentre registration network (www.ccgkdd.com.cn) that shares genotypes and phenotypes of inherited kidney diseases in children in China [15]. Patients with CAKUT were from 15 paediatric nephrology centres across China, where the physicians utilized the same standardized diagnostic and classification approaches after receiving standardized training from the ‘Internet Plus’ Nephrology Alliance of National Center for Children's Care. The Chigene database is a local Chinese database that contains whole exome sequencing (WES) data for ≈40 000 individuals (including 15 000+ trios) sequenced by the Chigene Translational Medical Research Centre (www.chigene.org). The Institutional Review Board of the Children's Hospital of Fudan University approved and monitored this study (no. 2018286).

Clinical diagnosis

Seven CAKUT categories were analysed. The first six were derived from one large study focusing on kidney outcome [12] as follows: (A) solitary kidney, (B) unilateral RHD, (C) bilateral RHD, (D) RHD associated with PUV, (E) multicystic dysplastic kidney (MCDK), (F) horseshoe kidney and (G) others. The ‘others’ included ectopic kidney and other outflow abnormalities [16], such as vesicoureteral reflux (VUR), pelvi-ureteric junction obstruction (PUJO), ureterovesical junction obstruction (UVJO), duplex collecting system and hydronephrosis, which are rare but may also affect kidney function.

The solitary kidney comprises kidney agenesis and the ‘empty renal fossa’ on imaging studies that may result from a multicystic kidney [12]. RHD is characterized by varying degrees of defective kidney formation. Only a kidney biopsy can help distinguish this, therefore renal hypoplasia and dysplasia were placed in the same category. MCDK was diagnosed by ultrasound evidence of a kidney with parenchyma completely substituted by large cysts of varying size, with an absence of function of that kidney determined by kidney scintigraphy [12]. Patients with symptoms indicating VUR, such as acute pyelonephritis, a history of recurrent or febrile urinary tract infection (UTI) or CKD of unknown aetiology, underwent a micturating cystourethrogram.

A prenatal diagnosis was made through a foetal ultrasound examination (oligohydramnios or variations in gross morphology of the kidney, ureter or bladder) [17]. Initial symptoms included UTI, gross haematuria, foamy or cloudy urine, abnormal urination (polyuria/nocturia, urge incontinence) and flank or abdominal pain. Asymptomatic patients may be detected incidentally when imaging methods are used for a different issue.

Kidney function was evaluated using the serum creatinine level and the estimated glomerular filtration rate (eGFR). eGFR was calculated using the updated Schwartz equation [18]. The severity of CKD was classified by the Kidney Disease: Improving Global Outcomes (KDIGO) guidelines [19]. Kidney failure was defined as GFR <15 ml/min/1.73 m² or treatment with kidney replacement therapy.

Patients with CKD stage 1 attended follow-up every 6 months, CKD stage 2–5 attended every 3 months and extrarenal manifestations were evaluated simultaneously.

Genetic sequencing and data analysis

Targeted exome sequencing (TES), proband-WES and trio-WES were performed for genetic testing. Low-coverage whole genome sequencing (WGS), multiplex ligation-dependent probe amplification (MLPA), chromosomal microarray analysis (CMA) and quantitative polymerase chain reaction were used for CNV analysis and identification.

Genomic deoxyribonucleic acid was isolated from blood lymphocytes. Exome sequencing, variant burden analysis and decision making regarding identified variants were performed using a previously reported strategy [15]. Exome capture was performed using the Agilent ClearSeq Inherited Disease Kit for TES or the Agilent SureSelect All Exon Human V5 Kit (Agilent, Santa Clara, CA, USA), xGen Exome Research Panel v1.0 (Integrated DNA Technologies, Coralville, IA, USA) or SeqCap EZ Exome Enrichment Kit v2.0 (Roche NimbleGen, Pleasanton, CA, USA) for exome sequencing, followed by next-generation sequencing using the Illumina HiSeq or Illumina NovaSeq 6000 sequencing platform (Illumina, San Diego, CA, USA). Sequence reads were mapped onto the human reference genome (NCBI build 37/hg19) with the Burrows–Wheeler Aligner (v.0.5.9-r16). The raw data generated by different platforms that met minimum quality standards were retained for further analysis, with mean coverage ≥20×, ≥90% target coverage region and having base call quality scores ≥Q20. Variant interpretation was performed using the same strategy (Supplemental Fig. S2) by nephrologists, molecular geneticists with domain expertise in inherited kidney diseases and bioinformaticians according to the American College of Medical Genetics (ACMG) guidelines [20]. Variants classified as pathogenic (P), likely pathogenic (LP) or variant of uncertain significance (VUS) were confirmed using Sanger sequencing. Candidate CNVs were annotated by analysing the genes contained in the CNVs and the CNV intervals themselves with the databases Decipher, ClinVar, ClinGen and Online Mendelian Inheritance in Man. CNVs overlapping known disease‐associated regions were detected according to the ACMG guidelines for CNV calling [21].

Genetic test methods included TES, proband-WES, proband-WES + CNV, trio-WES and trio-WES + CNV.

Statistical analysis

Continuous data are expressed as mean ± standard deviation (SD) or median and interquartile range (IQR). All categorical data are presented as numbers and percentages.

To determine differences between groups according to genetic diagnoses, categorical variables were analysed using the chi‐squared test or Fisher's exact test.

Statistical analyses of survival and multivariate analyses of factors associated with kidney survival were performed. The Kaplan–Meier survival curves were calculated and analysed using the univariate logrank test. Differences in the severity of the disease at presentation were considered, including sex, genetic diagnosis, CAKUT phenotypes and premature birth. The survival function and the median time to kidney failure were estimated by the Kaplan–Meier method. We used the date of birth as the starting point and the last point of the survival analysis was the age at progression to kidney failure or the last available observation. The patients who reached the last available follow-up without showing kidney failure were right-censored. The number of at-risk patients is listed by time point in the figures. The median outcome-free age was reported with a 95% confidence interval (CI).

Baseline variables considered clinically relevant or univariately related to outcomes were entered into a multivariate Cox proportional hazards regression model. Considering the number of events available, the variables included were carefully chosen to ensure the simplicity of the final model. No patient died before developing kidney failure (no competing risks). Final Cox proportional hazards analysis was adjusted for genetic diagnosis, CAKUT phenotype and premature birth. The baseline parameters were tested one by one for the proportional hazard assumption. The assumption validity was assessed by the log-minus-log survival function. For the assumption to hold, the log-log plot should show the separate lines as approximately parallel. No violation was observed. The association between predictors and outcomes in the Cox proportional regression model was expressed as hazard ratio (HR) and 95% CI.

All analyses were performed using SPSS version 19 statistical software (SPSS, Chicago, IL, USA) with default settings. P < .05 was set as the level of statistical significance.

RESULTS

Cohort characteristics

A total of 925 unrelated children with CAKUT were recruited; 584 patients were from the CCGKDD and had detailed clinical data available and 341 patients were from the Chigene database (Supplementary Fig. S1). Most cases were sporadic [n = 890 (96.2%)] and isolated to the urinary tract [n = 656 (70.9%)] (Table 1).

Table 1:

Baseline characteristics of 925 participants with CAKUT by genetic diagnosis.

	Genetic risk factors				Positive genetic diagnoses
Characteristics	All (N = 925)	Positive (n = 96)	Negative (n = 829)	P- value	Monogenic variants (n = 72)	CNVs and major chromosomal anomalies (n = 24)	P- value
Male	538 (58.2)	57 (59.4)	481 (58.0)	.276	41 (56.9)	16 (66.7)	.401
Positive family history	35 (3.8)	9 (9.4)	26 (3.1)	.007a	5 (6.9)	4 (16.7)	.221a
Genetic test methods				.015a			<.001a
TES	44 (4.8)	6 (6.3)	38 (4.6)		6 (8.3)	0 (0)
Proband-WES	157 (17.0)	11 (11.5)	146 (17.6)		11 (15.3)	0 (0)
Proband-WES + CNV	267 (28.9)	19 (19.8)	248 (29.9)		9 (12.5)	10 (41.7)
Trio-WES	343 (37.1)	40 (41.7)	303 (36.6)		39 (54.2)	1 (4.2)
Trio-WES + CNV	114 (12.3)	20 (20.8)	94 (11.3)		7 (9.7)	13 (54.2)
CAKUT subphenotypes				<.001a			.079a
Group A: solitary kidney	95 (10.3)	13 (13.5)	82 (9.9)		13 (18.1)	0 (0)
Group B: unilateral renal hypodysplasia	174 (18.8)	13 (13.5)	161 (19.4)		10 (13.9)	6 (25)
Group C: bilateral renal hypodysplasia	207 (19.1)	44 (45.8)	163 (19.7)		34 (47.2)	10 (41.7)
Group D: renal hypodysplasia associated with PUV	21 (2.3)	0 (0)	21 (2.5)		0 (0)	0 (0)
Group E: MCDK	79 (8.5)	3 (3.1)	76 (9.2)		0 (0)	0 (0)
Group F: horseshoe kidney	26 (2.8)	3 (3.1)	23 (2.8)		2 (2.8)	1 (4.2)
Group G: othersb	323 (34.9)	20 (20.8)	303 (36.6)		13 (18.1)	7 (29.2)
VUR	242 (26.2)	12 (12.5)	230 (27.7)		8 (11.1)	4 (16.7)
UVJO	3 (0.3)	0 (0)	3 (0.4)		0 (0)	0 (0)
Hydronephrosis	33 (3.6)	1 (1.0)	32 (3.9)		1 (1.4)	0 (0)
PUJO	4 (0.4)	1 (1.0)	3 (0.4)		0 (0)	1 (4.2)
Duplex collecting system	31 (3.4)	4 (4.2)	27 (3.3)		2 (2.8)	2 (8.3)
Ectopic kidney	10 (1.1)	2 (2.1)	8 (1.0)		2 (2.8)	0 (0)
Extrarenal anomalies c	269 (29.1)	62 (64.6)	207 (25.0)	<.001	48 (66.7)	14 (58.3)	.460
Behavioural and cognitive abnormalities	67 (7.2)	12 (12.5)	55 (6.6)	.036	6 (8.3)	6 (25)	.067a
Congenital heart disease	64 (6.9)	14 (14.6)	50 (6.0)	.002	9 (12.5)	5 (20.8)	.329a
Skeletal anomalies	54 (5.8)	10 (10.4)	44 (5.3)	.043	4 (5.6)	6 (25)	.014a
Genito-reproductive	34 (3.7)	6 (6.3)	28 (3.4)	.154a	4 (5.6)	2 (8.3)	.638a
Motor retardation	35 (3.8)	5 (5.2)	30 (3.6)	.39a	4 (5.6)	1 (4.2)	1.000a
Central nervous system	30 (3.2)	3 (3.1)	27 (3.3)	1.00a	1 (1.4)	2 (8.3)	.153a
Lung and diaphragm	2 (0.2)	1 (1.0)	1 (0.1)	.197a	1 (1.4)	0 (0)	1.000a
Face	23 (2.5)	10 (10.4)	13 (1.6)	<.001a	8 (11.1)	2 (8.3)	1.000a
Ocular	37 (4.0)	22 (22.9)	15 (1.8)	<.001a	20 (27.8)	2 (8.3)	.055a
Hearing	23 (2.5)	10 (10.4)	13 (1.6)	<.001a	9 (12.5)	1 (4.2)	.443a
External ear	28 (3.0)	11 (11.5)	17 (2.1)	<.001a	10 (13.9)	1 (4.2)	.281a
Gastrointestinal tract	27 (2.9)	8 (8.3)	19 (2.3)	.01	7 (9.7)	1 (4.2)	.675a
Diabetes mellitus	4 (0.4)	2 (2.1)	2 (0.2)	.05a	2 (2.8)	0	1.000a

Values presented as n (%).

^aFisher's exact test.

^bGroup G: others (VUR, UVJO, hydronephrosis, PUJO, duplex collecting system, ectopic kidney).

^cOne patient may have different extrarenal phenotypes.

Genotype–phenotype relationship in 925 patients with CAKUT

A genetic diagnosis was established for 96 of 925 (10.3%) children, including 72 (7.8%) with monogenic variants, 20 (2.2%) with CNVs and 4 (0.4%) with major chromosomal anomalies (Fig. 1, Table 1). Clinical phenotypes subdivided into seven CAKUT categories and variant types in each of the 96 genetically diagnosed patients are summarized in Supplementary Fig. S3. In total, 381 of 925 patients underwent a CNV analysis. The likelihood of establishing a genetic diagnosis was higher for patients who had a positive family history (25.7%) and extrarenal features (23.0%) (Table 1). For extrarenal manifestations, the two highest genetic diagnosis rates were ocular (59.5%) and hearing (43.5%) abnormalities (Table 1). Trio-WES with CNV analysis showed the highest genetic diagnostic rate (17.5%; Table 1). Although patients with skeletal abnormalities were more likely to have large CNVs or abnormal karyotypes than monogenic variants, the differences in the categories of genetic diagnosis in patients with and without extrarenal abnormalities were not statistically significant (Table 1).

Figure 1:

Genotype–phenotype relationship in 925 children with CAKUT. (A) The genetic spectrum varies among different CAKUT subphenotypes. (B) The percentage and number of patients in whom the genetic diagnosis was established in more than one patient. (C) The clinical diagnostic spectrum of patients with genetic diagnosis. BRHD, bilateral renal hypodysplasia; DS, duplex collecting system; EK, ectopic kidney; HD, hydronephrosis; HSK, horseshoe kidney; SK, solitary kidney; URHD, unilateral renal hypodysplasia.

The genetic spectrum varies among different CAKUT subphenotypes (Fig. 1A). A genetic diagnosis was identified in 13.7% of group A (solitary kidney), 7.5% of group B (unilateral RHD), 21.3% of group C (bilateral RHD), 3.8% of group E (MCDK) and 11.5% of group F (horseshoe kidney), respectively. A genetic diagnosis was not established for patients in group D with RHD associated with PUV and UVJO.

Figure 1B illustrates the percentage and number of patients in whom the genetic diagnosis was established in more than one patient. Pathogenic/likely pathogenic (P/LP) variants in PAX2 and 17q12 deletion classified according to ACMG guidelines were the top-two common causes.

Details of the genetic diagnosis are provided in Supplementary Table S1 (P/LP nucleotide variants) and Supplementary Table S2 (P/LP copy number variants). Two loci, chromosome 17q12 (all deletions) and chromosome 1q21.1 (all deletions), explained 10 of the 20 (50%) patients who carried a known CNV (Supplementary Table S3). Klinefelter syndrome (47,XXY) was the most common chromosomal abnormality (three patients). Eleven genetic disorders accounted for 66 of the 96 (68.9%) patients with a genetic diagnosis (Fig. 1B). The clinical diagnostic spectrum of patients with a genetic diagnosis is presented in Fig. 1C. Genotype–phenotype correlations determined that 17q12 loci were all identified in RHD and all COL4A1 variants were identified in VUR. Diagnostic variants of PAX2, GATA3, HNF1B, SALL1, KMT2D and TNXB were detected in different CAKUT subphenotypes.

Fifty-six identified variants were classified as VUS (Supplementary Table S3). In 2.1% of patients (19 of 925) we detected variants in 16 disease-causing genes known to cause monogenic non-CAKUT diseases (NPHP1, UMOD, WT1, EBP, ERCC8, DYNC2H1, RYR2, INSL3, SON, ERCC8, YY1, AVPR2, GATA6, WDR19, COL4A3 and DMD) (Supplementary Table S4). In these patients, no CAKUT-causing variants were identified.

Genotype–phenotype relationships in 584 children with CAKUT from the CCGKDD who had a detailed clinical history

Information on kidney function was available for 584 children from the CCGKDD. The characteristics are shown in Table 2 and Supplementary Fig. S4. For the 72 of 584 (12.3%) patients who had a genetic diagnosis, we analysed the detailed clinical data (the time point at which the first symptom was observed to the age at the last follow-up and the age at diagnosis of kidney failure) after subdivision into seven CAKUT subphenotypes to explore genotype–phenotype correlations (Supplementary Fig. S4). Overall, 128 patients (21.9%) were prenatally diagnosed and 287 patients (49.1%) were diagnosed after they presented with a UTI prompting imaging studies. The number of children with a family history [28 patients (4.8%)] or parental consanguinity [2 patients (0.3%)] was relatively limited.

Table 2:

Characteristics of 584 patients from the CCGKDD who had a detailed clinical history by CAKUT subphenotypes.

Category (n)	Gender, n (%)		Genetic diagnosis, n (%)	Genetic diagnosis with certain variantsa, n (%)	Premature birth, n (%)	CKD 5 at last follow-up, n (%)
	Female	Male
Group A (23): solitary kidney	9 (39.1)	14 (60.9)	4 (17.4)	4 (17.4)	1 (4.3)	7 (30.4)
Group B (112): unilateral renal hypodysplasia	18 (16.1)	94 (83.9)	13 (11.6)	7 (6.3)	7 (6.3)	10 (8.9)
Group C (135): bilateral renal hypodysplasia	47 (34.8)	88 (65.2)	38 (28.1)	25 (18.5)	22 (16.3)	57 (42.2)
Group D (19): renal hypodysplasia associated with PUV	0 (0)	19 (100)	0 (0)	0 (0)	6 (31.6)	4 (21.1)
Group E (7): MCDK	3 (42.9)	4 (57.1)	0 (0)	0 (0)	0 (0)	0 (0)
Group F (10): horseshoe kidney	6 (60.0)	4 (40.0)	1 (10)	0 (0)	1 (10.0)	0 (0)
Group G (278): othersb	139 (50.0)	139 (50.0)	16 (5.8)	1 (0.4)	10 (3.6)	4 (1.4)

^aGenetic diagnosis with certain variants: PAX2, TNXB, EYA1, HNF1B, GATA3 and 48,XXYY.

^bGroup G: others (vesicoureteral reflux, ureterovesical junction obstruction, hydronephrosis, pelvi-ureteric junction obstruction, duplex collecting system, ectopic kidney).

During a median follow-up of 4.7 years (IQR 1.4–7.4), kidney function was abnormal in 216 patients (37.0%). Eighty‐two (14.0%) patients progressed to kidney failure at a median age of 13.0 years (95% CI 12.4–13.6). Kidney survival probabilities were greater for categories E and F (no events) than for categories G (4), D (4), A (7), B (10) and C (57).

The patients with genetic disorders showed worse outcomes (Fig. 2A). Of the 82 patients who progressed to kidney failure, 24 were genetically diagnosed with P/LP variants of PAX2, TNXB, EYA1, HNF1B and GATA3 or the 48,XXYY karyotype (Supplementary Fig. S4). Thirty-seven patients had the above variants and they had worse outcomes than the other 547 patients (χ² = 18.01, P < .001) (Fig. 2B). The median age at the diagnosis of kidney failure was 10.0 years (95% CI 6.0–14.0) for patients with certain variants described above and 13.0 years (95% CI 11.5–14.5) for others.

Figure 2:

Kaplan‒Meier kidney survival curves for 584 children with CAKUT from the CCGKDD stratified according to (A) categories of genetic diagnosis; (B) the presence of certain variants (PAX2, TNXB, EYA1, HNF1B, GATA3 and 48,XXYY); (C) sex and (D) preterm or term birth.

Sex did not influence the outcome (Fig. 2C). The prognosis was significantly poorer for patients with premature birth (Fig. 2D).

Adjusted survival curves based on the Cox regression model

The multivariate Cox regression analysis showed solitary kidney; PUV; bilateral RHD; P/LP variants of PAX2, TNXB, EYA1, HNF1B and GATA3 or the 48,XXYY karyotype; and premature birth were independent risk factors for kidney failure (Table 3). A solitary kidney, PUV and bilateral RHD predicted worse outcomes independent of other covariates. After controlling for certain genetic diagnoses and premature birth, a significantly different clinical course was observed in CAKUT subphenotypes (Fig. 3).

Table 3:

Univariable/multivariable Cox regression analysis of probable explaining factors for progression of kidney failure in children with CAKUT from the CCGKDD.

	Univariable analysis			Multivariable analysis
Variable	HR	95% CI	P-value	HR	95% CI	P-value
Genetic diagnosis with certain variantsa	2.754	1.686–4.497	<.001	1.742	1.060–2.864	.029
CAKUT categoryb			<.001			<.001
Posterior urethral valves	18.550	4.631–74.309	<.001	12.185	2.942–50.461	.001
Solitary kidney	11.546	3.356–39.722	<.001	9.123	2.600–32.010	.001
Bilateral hypodysplasia	12.859	4.602–35.932	<.001	10.042	3.544–28.455	<.001
Unilateral hypodysplasia	3.524	1.081–11.493	.037	3.032	0.924–9.949	.067
MCDK, horseshoe kidney, and othersc	–	–	–	–	–	–
Premature birth	3.766	2.099–6.758	<.001	2.844	1.531–5.282	.001

^aGenetic diagnosis with pathogenic/likely pathogenic variants of PAX2, TNXB, EYA1, HNF1B, GATA3 and 48,XXYY.

^bThe reference level for the effect estimates associated with the CAKUT categories (solitary kidney, bilateral hypodysplasia, unilateral hypodysplasia and posterior urethral valves) includes multicystic kidney, horseshoe kidney and others.

^cOthers: vesicoureteral reflux, ureterovesical junction obstruction, hydronephrosis, pelvi-ureteric junction obstruction, duplex collecting system and ectopic kidney.

Figure 3:

Survival probabilities of 584 children with CAKUT from the CCGKDD stratified by CAKUT subphenotypes after adjustment for the presence of certain variants (PAX2, TNXB, EYA1, HNF1B, GATA3 and 48,XXYY) and premature birth.

DISCUSSION

In this study, an in-depth analysis of the phenotypes and genotypes of 925 children with CAKUT was performed. The genetic spectrum varies among different CAKUT subphenotypes. To the best of our knowledge, this multivariate analysis is the first to include the genetic diagnosis and detailed clinical phenotype in a large cohort of children with CAKUT for the risk factors for kidney failure.

A genetic diagnosis was established for 10.3% of the children. The genetic diagnostic rate (21.3%) was higher for patients with bilateral RHD. Although the rate was relatively low (5.0%), genetic diagnoses were detected in 12 of 242 patients with VUR. The same diagnostic genes were obtained for different CAKUT subphenotypes. P/LP variants in PAX2 and 17q12 were common causes, similar to previous findings [22, 23].

To date, 50 different monogenic causes [24] and 6 CNVs (17q12, 22q11.2, 1q21, 16p11.2, 4p16.1-p16.3, 16p11.3) [25] have been described as major causes of CAKUT. Table 4 and Supplementary Table S5 summarize the molecular genetic studies (>60 participants) of CAKUT and show the variability in the detection rate of genetic causes due to the differences in methodology, study design and population [9, 10, 22, 23, 26–30]. Literature also reinforces our results, suggesting that genetic analysis of CAKUT genes is highly recommended for patients with a family history and syndromic CAKUT [31] and necessitates surveillance for the possible clinical features associated with genetic defects [6, 22, 23]. In this study, CNV analysis and karyotype testing increased the diagnostic yield of genetic testing by 2.6%. Patients, especially those with skeletal abnormalities, may benefit more from a CNV analysis and karyotyping.

Table 4:

Summary of molecular genetic studies (>60 participants) of CAKUT.

ID	Year	Country	Family (patients)	Positive familial history, %	Sex (male), %	Extrarenal phenotypes, %	CAKUT cohort	CAKUT exclusion	2–3 diagnoses, %	Methods	Gene number	Causative variants, %	Causative CNV, %	Causative variants (patients)	Reference
1	2006	European	99 (100)	12	59	23	RHD	RHD with PUV or primary bladder abnormalities	NA	TES	5	20	ND	TCF2 (8), PAX2 (7), EYA1 (2), SIX1 (2), SALL1 (1)	25
2	2014	World-wide	650 (749)	21	55	0	Isolated CAKUT (288 VUR, 120 RHD, 90 solitary kidney)	Syndromic CAKUT	21	TES	12	6	ND	BMP7 (1), CDC5L (1), CHD1L (5), EYA1 (3), GATA3 (2), HNF1B (6), PAX2 (5), RET (3), ROBO2 (4), SALL1 (9), SIX2 (1), SIX5 (1)	24
3	2016	Worldwide	453	10	61	14	132 duplex collecting system, 103 PUJO, 60 PUV, 53 VUR	NA	43	TES	208	1	ND	PAX2 (3), HNF1B (1), UMOD (1), SIX5 (1)	26
4	2017	France	204	30	57	39	Bilateral or unilateral associated with either extrarenal defects or familial history	PUV	35	TES	330	18	ND	HNF1B (9), PAX2 (9), EYA1 (5), GATA3 (4), ANOS1 (2), CHD7 (1), KIF14 (2), PBX1 (4)	27
5	2017	USA	62 (112)	16	56	31	Isolated and syndromic without a known diagnosis, 23% RHD, 19% solitary kidney, 16% PUV, 15% VUR	Syndromic CAKUT in which an underlying genetic aetiology was known and individuals with non-syndromic and non-familial forms of VUR	NA	WES (20 trio-WES)	NA	5	6.50	EYA1 (1), HNF1B (1), PAX2 (1), FOXP1 (1)	28
6	2018	Worldwide	232 (319)	17	59	25	CAKUT and cryptorchidism	NA	17	WES	404	14	NA	22 different monogenic genes	21
7	2019	Japan	66	NA	76	41	Bilateral renal lesions, extrarenal complications or a family history	VUR alone and chromosome aberrations	NA	TES, MLPA, aCGH, NGS	NA	21	6	HNF1B (7), PAX2 (4), EYA1 (1), CHD7 (1), EP300 (1)	8
8	2020	Korea	94	5	83	34	Sever CAKUT (solitary kidney, RHD, MCDK, bilateral VUR and PUV)	Unilateral VUR, unilateral obstructive uropathy	29	TES	60	7	6.40	HNF1B (2), PAX2 (2), EYA1 (1), UPK3A (1), FRAS1 (1)	9
9	2022	China	925	4	58	29	CAKUT	NA	31	TES, WES, low-coverage WGS, MLPA, aCGH	174	8	2	This study	This study

^aCGH, array comparative genomic hybridization; MPLA, multiplex ligation-dependent probe amplification.

This study illustrates the association between genotypes and other risk factors for kidney failure in a multicentre paediatric cohort with a wide variety of CAKUT subphenotypes utilizing the same diagnostic and classification approaches. Accurately stratifying patients with CAKUT at risk for progression could enable earlier targeted treatment to stabilize kidney decline and reduce future adverse outcomes.

A study of 66 patients with CAKUT found the kidney prognosis varied for each genetic variant; 10-year kidney survival was 86% for the HNF1B variant group, 38% for the PAX2 variant group and 100% for the other gene variant group (P = .08) [9]. A study of 94 patients revealed that patients with variants of HNF1B, PAX2, EYA1 and FRAS1 progressed to kidney failure [10]. Our previous study of 379 children with primary VUR revealed that PAX2 or TNXB variant carriers had worse outcomes [13]. A recent study found HNF1B and PAX2 were responsible for the majority of diagnoses in CAKUT patients with early-onset CKD [32]. These findings all indicate the necessity of classifying different genes. Therefore we analysed the genes that have been reported or progressed to kidney failure in this cohort rather than all molecular diagnoses. Patients carrying PAX2, TNXB, EYA1, HNF1B or GATA3 variants or with the 48,XXYY karyotype had a shorter predicted kidney survival. Knowing the differences in the kidney prognosis related to the genetic variants is beneficial because these gene variants generally have a high detection rate.

Bilateral RHD was reported to be associated with a higher risk of kidney failure [12, 33]. CAKUT subphenotypes have a more significant impact on prognosis. Survival data stratified by CAKUT subgroups showed a significantly poorer outcome for patients assigned to the solitary kidney, PUV and RHD groups, consistent with published data [12]. Isolated VUR and other urinary tract anomalies, such as duplicated collecting systems, represented a relatively good prognosis.

Another important prognostic indicator for progression to kidney failure is premature birth, since extreme prematurity has been shown to correlate with low nephron number in autopsy studies [34]. The univariate analysis by Isert et al. [11] produced similar results, showing prematurity is associated with CKD stage ≥2.

This study has several limitations. First, the family chose whether genetic testing was performed, and triplicate samples were not available for every family. Because the pathogenicity must be further classified, VUS was categorized as no genetic diagnosis, which may lead to the underestimation of genetic data. Second, our cohort had a higher proportion of patients with kidney failure due to the focus of the Nephrology Alliance of National Center, and the selection of certain genetic diagnoses as a composite predictor variable is exploratory, which may affect the results applicable to the general population. Finally, our findings are potentially limited by other factors, such as inhibition of the renin–angiotensin system, management of hypertension and surgical intervention.

In conclusion, the genetic spectrum varies among different CAKUT subphenotypes. Patients with skeletal abnormalities may benefit more from a CNV analysis and karyotyping. Patients with PAX2, TNXB, EYA1, HNF1B or GATA3 variants or the 48,XXYY karyotype had a shorter predicted kidney survival. A solitary kidney, PUV, bilateral hypodysplasia and premature birth are associated with an increased risk of kidney failure during childhood. Close follow-up visits in accordance with the identified factors are recommended. Prospective studies are needed to investigate the applicability of these results.

FUNDING

This work was supported by the National Key R&D Program of China (2021YFC2500202) and National Natural Science Foundation of China (NSFC-81873593, NSFC81800602, NSFC82070686), a grant (20184Y0176) from the Shanghai Municipal Commission of Health and Family Planning Youth Research Program, a grant (SHDC2020CR2064B) from the Clinical Research Plan of Shanghai Hospital Development Center and a grant (EK2022ZX01) from the Key Development Program of Children's Hospital of Fudan University.

AUTHORS’ CONTRIBUTIONS

H.X. and Q.S. were responsible for the concept and management. J.L.L., X.W.W. and C.H.L. participated in the execution and coordination. J.L.L. drafted the manuscript. J.L.L., X.W.W., C.H.L., D.M., X.J.G., X.Y.J., J.H.M., G.H.Z., A.H.Z., M.W., X.Q.D., J.Q.Z., Y.F.L., H.T.B., R.F.Z., T.S., Y.L.B., Y.B.S., X.W., B.B.W., J.C., J.R. and X.S.T. conducted data acquisition, analysis and interpretation.

DATA AVAILABILITY STATEMENT

The data underlying this article are available in the article and in its online supplementary material.

CONFLICT OF INTEREST STATEMENT

The authors have no conflict of interest to declare.