Background
Steroid resistant nephrotic syndrome (SRNS) remains one of the most common intractable causes of end-stage renal disease (ESRD) in children with 50–70% of these children developing end-stage renal disease within 5–10 years of diagnosis [
1]. The therapeutic options in SRNS are often inefficient, and complicated by significant toxicity adding to the associated morbidities, mortality and cost. There is now compelling evidence that children with pathogenic variations in the genes responsible for maintenance of podocyte structure and function form a distinct subgroup of Nephrotic Syndrome (NS) and these children are generally unresponsive to immunosuppression, but do not have post-transplantation recurrence [
2,
3].
More than 53 single gene mutations specific to podocyte or associated with glomerular filtration barrier have been found to be associated with SRNS [
4,
5]. Large multi centric studies including population of multiple ethnicities showed genetic mutations in about ~ 30% of SRNS patients with a higher proportion in infants and young children. Most mutations were observed in
NPHS2, WT1 and
NPHS1 genes [
4,
6].
However, reports from India including from our center showed that the prevalence of
NPHS2 mutations is much lower in Indian population when compared with Europe and North American population [4% vs. 10.5–28%)] [
7‐
12]. Kumar et al., reported low prevalence of
WT1 mutation in south Indian population, whereas we did not detect any mutation in
WT1 gene in 100 SRNS children [
13,
14]. These data suggest that a traditional genetic testing using an algorithmic approach based on age of onset of NS to prioritize the genes to be sequenced by Sanger may not be useful [
15,
16]. The above data also indicates the need for additional screening of genes implicated in SRNS in order to understand the genetic spectrum of SRNS in Indian population. Given the genetic heterogeneity and phenotypic variability in SRNS, Sanger sequencing is not a feasible approach for routine testing. Next-generation sequencing (NGS) technology is emerging as a cost-effective strategy to screen multiple genes in genetically heterogeneous diseases like SRNS [
17].
The aim of our study was to check the feasibility of genetic diagnosis using targeted next-generation sequencing (NGS) approach in Indian children with SRNS. We report the initial results along with the challenges faced in the analysis and interpretation of sequencing data obtained by simultaneously sequencing 17 genes in 25 children with SRNS and 3 healthy individuals.
Discussion
Identifying the cause of SRNS is of great importance as it helps in preventing unnecessary exposure to immunosuppressants and their adverse effects, besides establishing a molecular diagnosis and clear prognosis. It also enables targeted treatment as in case of children with pathogenic variants identified in gene encoding enzymes of the co-enzyme Q 10 biosynthesis who are amenable to treatment with coenzyme Q 10 [
32].
We report the results of sequencing for molecular diagnosis of SRNS in Indian children by screening 17 genes wherein pathogenic variant in NPHS2 gene was identified in 8% patients. Siblings carrying this variant along with the patient 83 carrying the NPHS2 variant H141Y were included as positive samples to check the sensitivity of the present assay. Both these variants were detected (although variant H141Y was initially missed due to low read depth) and no spurious pathogenic mutations were found in any of these samples indicating 85% sensitivity for the assay. Beside these known variants, 3 novel likely pathogenic variants were identified in 3 patients (12%) who were previously sequenced for NPHS2 and WT1 genes. These findings demonstrate the utility of NGS in a clinical setting since it allows for rapid and simultaneous screening of multiple SRNS associated genes instead of prioritizing specific genes for genetic testing.
The targeted gene panel was developed based on the results from two largest SRNS cohorts one of which included Indian children with SRNS. The targeted panel included 17 genes which explained the genetic basis in > 95% of children with SRNS in these two cohorts. Previous studies using the targeted multi-gene sequencing to analyze the exon and intron boundaries of genes associated with SRNS in various populations identified mutations in ~ 30% of the patients [
4‐
6,
21,
33‐
37]. In the present study, disease causing variants were identified in 20% of the cohort which is lesser than that expected probably due to small number of patients included in the cohort.
The most common disease causing variants were identified in the
NPHS2,
WT1, and
NPHS1 genes in the Podonet cohort (1174 patients from 21 countries; included 9 Indian patients = 0.7%), in 1783 unrelated, multinational cohort and in the UK cohort [
21]. This in contrast to the Chinese population, wherein the disease causing variants were also identified in
ADCK4 gene (6.67%), in addition to
NPHS1,
WT1, and
NPHS2 genes [
37]. In the present study, although the cohort size was small, disease causing variants were identified in
NPHS2 (12%)
NPHS1 (4%) and
PLCe1 (4%) genes indicating that the genes with variants causing SRNS varies significantly according to ethnic background. While this study and our previous study indicate that
NPHS2 gene is the most common mutated gene in Indian population [
7], we also identified
NPHS1 and
PLCe1 genes mutations that would not have been considered in the conventional genetic testing algorithms for SRNS using Sanger sequencing.
All the pathogenic variants were identified in genes associated with recessive Mendelian inheritance, as most of the children (64%) in the cohort developed SRNS at an early age (< 5 years). The age of onset in our study correlated with risk for an as reported in other studies [
6,
16]. Surprisingly, we did not find any pathogenic variants in infantile group. This is contrast to the findings from other studies where in ~ 66.3% of SRNS cases (onset between 0 and 1 year) is due to the mutation in one of following four genes:
NPHS1, NPHS2, LAMB2, or WT1 [
38]. This indicates that additional SRNS associated genes needs to be screened in this group.
It is well known that SRNS exhibits significant inter and intra familial variability. The use of NGS allows to study the influence of disease causing variants in multiple genes on phenotype variability [
33]. In the present cohort, two siblings with identical pathogenic variant (
NPHS2 R71X; SRNS20 and SRNS76) showed different clinical course. The variability in the clinical phenotype of patients carrying the same variant indicate an environmental factor or a possible second-site genetic modification, whereby pathogenic variants in a second gene might modulate the penetrance and/or expressivity of recessive mutations in a primary locus. Although in the siblings we did identify additional variant (R408Q) in
NPHS1, it was heterozygous and classified as begnin by both ACMG and Karbassi et al. variant scoring system [
25,
27]. In patient 83, two variants in the
NPHS2 gene (splice site, g.179521737C > T and missense H141Y) were identified. The splice site variant was classified as likely pathogenic while the H141Y variant was classified as VUS, with the risk score suggesting pathogenic nature. It is difficult to predict which variant is contributing to the disease development in this child. In order determine the role of multiple variants on the phenotypic variability we need to compare patients with different genotype combinations in the various cohorts that have been studied.
The main barrier to determine the pathogenicity of a variant is absence or limited functional testing of variants discovered to identify specific variants that results in dysfunction of the protein product. For example, a novel homozygous variant R752X, in PLCe1 gene in patient 123 was classified as likely pathogenic instead of pathogenic. Based on the clinical findings and histopathology of patient 123, it is evident that PLCe1 gene variant can potentially be attributed to the disease development in this patient. However, lack of data which would help with the segregation of alleles in cases and the reference population and absence of functional data, we were unable to classify this variant as pathogenic.
Secondly, guidelines to annotate the heterozygous variants in dominant genes are not very clear. For example the novel LMX1B gene variant V145 M with low allele frequency was predicted to be pathogenic in nature as per the Karbassi scoring algorithm but still classified as VUS as per the ACMG criteria. Further functional studies are required to confirm the effect of this variant on protein function and disease phenotype. Since little robust data is available upon which to base an assessment of causality in case of VUS, reporting, genetic and medical counseling can be complex and challenging. There is no consensus on optimal strategies to report such findings and for clinician to communicate them with parents. Counselling parents with an affected child with a VUS is even more challenging in a prenatal setting as quantifying the attributable risk of developing the disease is not possible if the variant is prospectively detected in the unborn fetus. Hence developing appropriate and effective clinical approaches to this challenge including additional training to clinicians in pretest counseling and consenting, interpretation of results and communication of results to the parents is essential. Besides, integrating the data from this study with large publically accessible phenotype and genotype data may help in ascertaining the role of novel variants in disease development and also determine the role of multiple variants on the phenotypic variability.
This study is unique as it is the first Indian study using well phenotyped SRNS cohort and NGS technology for the genetic diagnosis of SRNS. However it had few limitations such as non-random sample selection (majority of the patients were early childhood onset) and selection of small number of patients from a single center. As parental DNA was not available we could not perform segregation studies in the familial cases except in one family.