Differential methylation as a diagnostic biomarker of rare renal diseases: a systematic review

Whilst rare diseases are uncommon at the individual level, cumulatively they represent a significant public health problem with approximately 350 million people suffering worldwide [1]. The definition of a rare disease varies between continents; the European Commission classifies a rare disease as one which affects less than five people in 10,000, whilst the American definition states rare diseases affect fewer than 200,000 people [1]. An underlying genetic cause is suspected in 80% of rare conditions, with 50% of these diseases occurring in children. Two in five patients with rare diseases describe struggling to obtain a timely accurate diagnosis which can be detrimental to each individual’s quality and length of life [2]. Interpretation of massive amounts of genetic information generated by large scale sequencing efforts remains a challenge; but despite this, these efforts are significantly improving the speed and accuracy of rare disease diagnosis [3]. Additionally, rationalising data generated by multi-omic approaches could provide new insights into molecular profiles for rare diseases.

More than 100 rare renal disorders have been reported [4, 5]. Similar to many other rare diseases, clinical diagnosis of rare renal diseases may be challenging with limited clinically relevant biomarkers, significant phenotypic variability, poor understanding of the disease pathogenesis and a lack of appropriate therapies. For example, diagnosis of IgA nephropathy (IgAN) involves a renal biopsy to confirm diagnosis which is an invasive and relatively expensive procedure [6]; diagnosis of polycystic kidney disease (PKD) utilises ultrasound scanning for diagnosis, which is insufficiently sensitive to detect the earliest stages of PKD even though earlier detection may improve the patient’s prognosis [7]. The development of a panel of cost-efficient, sensitive and accurate biomarkers of rare kidney disease which can be detected non-invasively would greatly aid rare renal disease diagnosis. A biomarker is defined as “a characteristic that is objectively measured and evaluated as an indicator of normal biological processes, pathogenic processes, or pharmacologic responses to a therapeutic intervention” [8]. At present, biomarkers to identify renal dysfunction are sub-optimal and often based on inexact biochemical markers where half of ‘normal’ kidney function is lost before kidney disease is identified. Additionally it is important to note that kidney dysfunction (associated with gene variants or complex chromosome abnormalities) can have widespread impacts on other organ systems [5]. Therefore, multi-centre studies are required to establish disease-based cohorts for rare renal disease with consistent biological sample collection, harmonised biomarker measurements, effective data sharing, and compatible analytics across all data.

Unravelling the genetic basis of rare kidney disorders has been facilitated by the establishment of rare renal disease biobanks and registries such as The PodoNet Registry for congenital and steroid resistant nephrotic syndrome [9], the UK National Registry of Rare Kidney Diseases (RaDaR) [10], and the development of focused international professional multidisciplinary teams such as the ERA-EDTA Working Group on Inherited Kidney Disorders [11]. Current research to improve knowledge of disease is moving beyond simple changes to the DNA sequence to utilising integrated multi-omic strategies to improve clinical diagnosis [12]. There are 517 omics currently described [13], with transcriptomics (RNA-based), epigenomics (DNA and RNA based non sequence level modifications), and proteomics being the primary research tools for rare renal diseases at present. Integrated molecular and clinical approaches are increasingly being employed to provide functional evidence for pathogenicity of SNPs (single nucleotide polymorphisms) and to deliver novel disease insights [14‐17].

Methylation, the addition of a chemical methyl group via DNA methyl transferases, is a key epigenomic feature affecting gene expression. Methylation predominately occurs in CpG dinucleotides (though it can occur elsewhere) where the methyl group is added to the fifth carbon of the cytosine forming 5-methylcytosine (5mc) [18]. Large-scale methylome-wide studies have allowed better understanding of DNA methylation and health-related outcomes, for example using Illumina’s Infinium MethylationEPIC array, which quantitatively and cost effectively interrogates approximately 850,000 features [19], or deep whole-genome bisulfite sequencing of genomic DNA [20]. Evidence is growing that methylation risk scores (analogous to genetic risk scores) may be constructed for multiple health outcomes [21]. Methylation is largely considered be a transcriptional repressor with roles in genomic imprinting [22], X Chromosome inactivation [23], repression of repetitive elements [24], aging [25] and tissue specific gene expression [26]. The location of the methylated sites affects function, for example methylation within the gene body rather than the classically thought of transcriptional start sites may have a role in transcriptional activation [27].

Aberrant methylation is implicated in multiple disorders affecting a range of organ systems. These include, but are not limited to, vascular complications in type 2 diabetes despite good glycaemic control [28], several cancers [29], clinical heterogeneity in Alzheimer’s disease [30], pre-natal and early childhood neurodevelopmental disease [31], development of chronic kidney disease [32, 33], and differential methylation has even been implicated by a number of studies as potentially impacting kidney transplant outcomes through alloimmune response and ischemia–reperfusion injury [34]. The crucial role of DNA methylation in monoallelic imprinting is now evident from disorders of methylation, when aberrant methylation has detrimental effects on development, as is often the case in Prader–Willi syndrome, Silver–Russell syndrome, Beckwith–Wiedemann syndrome and Type Ib pseudohypoparathyroidism [35, 36]. The sophistication of epigenetic tools for disease characterisation continues to improve, such as the development of the EpiSign assay which can aid the diagnosis of 19 often difficult to identify disorders, including Angelman Syndrome, Prader-Willi syndrome and Beckwith-Wiedemann syndrome [37]. DNA methylation can be influenced by inherited (genetic) and acquired throughout life (somatic; environmental) factors, creating changes that may be short-acting, exist long-term in an individual, and / or demonstrate transgenerational inheritance. Several recent multi-centre papers confirm socioeconomic experiences across the life course impact peripheral blood-derived methylation, persisting from pre-birth, throughout childhood, to later adult life [38‐41].

This review was designed using the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) systematic review checklist (Additional file 1) [42].

Sources initially identified from each database were as follows, MEDLINE n = 58, EMBASE n = 136, PubMed n = 62, Cochrane Library n = 1, GreyLit n = 0, OpenGrey n = 0. Following title and abstract screening, 94 duplicates were removed and 15 papers were identified for full text screening. No further studies were identified from searching genomeweb and Epigenesys publication databases. Finally, 13 texts were included in the full review, (Fig. 1), with the characteristics of each source summarised from the completed data extraction forms, (Additional file 2: Table S4). This included study aims, design information, methylation measurement, methodological rigour and key results. Outcomes, methylation measurement and participant information are briefly summarised (Table 1). Of the texts included, 12 were case-control studies and one was a case report. Reference lists of these papers and forward citations were also screened but no further sources were identified. Of the 13 articles included in this review, three originated from South Korea, two from China, one from Italy, one from Canada, three from the United States of America and three from Japan. Methodological rigour was assessed as weak for all studies, primarily based on lack of description dealing with limitations, such as regression to identify confounding factors, lack of appropriate matching of cases to controls on the basis of gender, age or ethnicity, and often a lack of detail on experimental controls, statistical analysis or animal model strains used.

After systematically evaluating current publications relevant to differential methylation in patients with rare renal diseases, this review has identified limited evidence for differential methylation in rare renal diseases. However, what evidence exists is promising and highlights the need for further research to explore differential methylation as a diagnostic and / or prognostic biomarker of rare renal diseases. DNA methylation and renal cancer is reviewed extensively elsewhere and is not discussed in this review [44].

At present, although rare forms of kidney disease significantly affect individuals living with these conditions, there is very poor understanding of the molecular characteristics and best treatment options for these conditions. Even where the underlying genetic cause is known for a rare renal disease there may still be significant unexplained heterogeneity in phenotypes between individuals with the same genotype. Therefore, studying epigenetic features, such as DNA methylation, may offer new insights by providing a mechanism to understand how each individual’s genome interacts with their environment through the epigenome. However, this review highlights that only a small number of studies have been reported researching differential methylation in rare renal diseases:

This reflects only 5% of the approximated total number of rare renal diseases which exist and thus a significant gap in published research has been identified [4, 5].

IgAN is a condition wherein Immunoglobulin A accumulates in kidney tissues and results in harmful inflammation, which can ultimately lead to end stage renal disease (ESRD) requiring dialysis and renal replacement therapy (RRT). IgAN was a disease included in the RaDaR recruitment criteria and familial IgAN was included in the 100,000 Genomes Project eligibility criteria. In the three studies of IgAN included in this review, DNA methylation was measured through chromatin immunoprecipitation (ChIP) microarray, whole genome microarray analysis of bisulphite converted DNA and methylation specific polymerase chain reaction (MSP). Aberrant methylation was identified as a potential driver in IgAN pathogenesis when identified in CD4+ T cells by causing T helper cell imbalances [49], as a contributor to abnormal glycosylation of IgA1 in IgAN through differential methylation of Cosmc [48], and as alterations in the H3K4me3 status identified in IgAN patients along with three significantly differentially methylated candidate genes (FCRL4, IL1RAPL1 and PTPRN2) that may exacerbate IgAN pathology through mediating the cytokine/chemokine cascade and inhibition of protein tyrosine kinase respectively [50].

ADPKD, included in the 100,000 Genomes Project and RaDaR eligibility criteria, is a disease where patients suffer from the growth of cyst formation on the kidney can lead to ESRD and other complications such as kidney stones, polycystic liver disease and brain aneurysms. Our systematic review identified three studies of DNA methylation and ADPKD, where methylation was measured through ChIP and methylated-CpG island recovery assays (MIRA-Seq). These studies identified differential methylation to have a role in ADPKD via upregulation of SMYD2 contributing to renal cyst formation due to methylation of STAT3 and p65, subsequently resulting in increased renal cell proliferation [53], epigenetic silencing of PKD1 and other ion transport genes in ADPKD due to hypermethylation [52], and identification of reduced expression of MUPCDH as a prognostic biomarker of ADPKD [51]. Of interest, the latter two of three studies noted that treatment with methylation inhibitor alleviated cyst formation, thus identifying novel therapeutic targets for ADPKD.

Six of the included articles discussed rare causes of proteinuria, including membranous nephropathy and FSGS (FSGS is included in the RaDaR recruitment criteria). Membranous nephropathy is a rare auto-immune glomerular disease with a global average incidence of 2.5/100,000 individuals [61], characterised by thickening of the glomerular wall and decreased filtration, leading to proteinuria and ultimately loss of kidney function. Our review identified two articles of differential methylation in membranous nephropathy, measured by ChIP sequencing and assays. Increased H3K4me3 was found to exacerbate proteinuria in membranous nephropathy [55], with murine models showing that targeting shRNA against an H3K4 methyltransferase, MLL3, alleviated proteinuria. H3K9me3 alterations were also found to be a biomarker of membranous nephropathy compared to normal control patients [54]. FSGS, included in the RaDaR elgibility criteria, is a condition describing sclerosis of the kidney, with a variety of causes. Three studies of proteinuria featuring FSGS and differential methylation were included, measured by whole genome microarray, MSP and ChIP assays. These studies identified that aberrant methylation in FSGS may cause proteinuria by downregulation of KLF4, which has a role in reprogramming somatic cells into induced pluripotent stem (iPS) cells [57]. This was further investigated to show that downregulation of KLF4 causes nephrin promoter methylation leading to development of proteinuria, which can be alleviated by treatment with an angiotensin receptor blocker (ARB), thus identifying a potential therapeutic target [56]. Interestingly, nephrin was also found to regulate epigenetically in proteinuria caused by FSGS, through aberrant methylation of WHSC1L1-L through interaction with H3K4 and H3K36 [58]. Downregulation of H3K27me3 causing subsequent upregulation of the Notch pathway was found to be associated with albuminuria in glomerular disease, including FSGS [59].

Finally, our comprehensive search identified one study of differential methylation in congenital renal agenesis [60]. Congenital renal agenesis is defined as a condition where one or both kidneys are missing at birth. In this case report of discordant monozygotic twins with congenital renal agenesis, differentially methylated regions are identified but location and whether this is an increase/decrease in methylation is not specified, as well as no discussion of the potential function of these differentially methylated regions.

A number of the studies included in this review investigated methylation in animal models and cell lines, which are imperfect models of human kidney disease [53, 58]. Therefore, it would be helpful to also analyse these candidate differentially methylated genes using DNA isolated from kidney biopsy tissue, saliva, urine or blood in patients with rare renal diseases to assess their utility as diagnostic biomarkers from a minimally invasive test that could be routinely performed in clinic or at a GP surgery; this would minimise risk to patients and improve cost-effectiveness for healthcare professionals.

Despite the scope of this review including literature from 1946, the research identified and described in this review has been conducted within the last decade, reflecting growing awareness of the potential utility of epigenetics in renal disease. Whilst this is encouraging, the studies measured differential methylation in fundamentally different ways and as such comparison between the outcomes were difficult. These different approaches to research methylation were as follows:

The final objective of this review was to evaluate any research which highlighted the potential for differential methylation as a diagnostic biomarker of rare renal disease. Only two of the studies discussed this potential biomarker development, one being the potential of methylation as a diagnostic biomarker for membranous nephropathy [54] and one the potential for a prognostic biomarker of ADPKD [51]. The focus of the majority of the articles was elucidating pathogenic mechanisms or developing a novel therapies. Therefore, this review highlights that although limited evidence does exist for differential methylation influencing rare renal diseases, further research is required to robustly identify differentially methylated features which could be potential biomarkers and to perform validation studies on such features.

The scarce nature of rare diseases makes it difficult to perform “gold standard” experimental design studies, such as an ethical randomised control trial with strong methodological rigour. One consistent limitation seen across studies included in this review was a lack of accounting for confounding factors in all case-control studies, even though such factors were frequently identified in the discussion and regression analysis may have been appropriate to improve the methodological rigour. This is typical for a developing field, therefore future studies would benefit from standardisation of the detection and analysis of methylation, the inclusion of more robust (laboratory and computational) quality controls, and a comprehensive, transparent reporting structure such as exists for genetic association studies [62].

The wide range of countries from which this research has originated is encouraging since international collaboration is essential to maximise the power of any study given the scarcity of participants with rare renal diseases. Projects such as UK’s 100,000 Genomes Project have left a lasting legacy with ongoing multi-omic analysis (including DNA methylation) currently helping extend understanding and knowledge of rare renal conditions [43]. Additional research would also be helpful to explore if there are significant differences associated with methylation and rare renal diseases between different ethnic groups.

This review highlights that there has been limited investigation of differential methylation for rare renal diseases, but this limited research is encouraging and will help guide future studies. For example, differential methylation of membranous nephropathy cases compared to healthy controls, significant hypermethylation and under expression of mucin-like protocadherin (MUPCDH) between ADPKD kidney tissue and non-ADPKD kidney tissue, differentially methylated regions in congenital renal agenesis, and significantly higher DNA methylation in genes FCRL4, PTPRN2 and IL1RAPL1 of IgAN patients compared to healthy controls; all of which highlight DNA methylation as a potential novel biomarker of rare renal disease. Further research focus is required for standardised, international multi-omic analysis of rare renal diseases towards developing a panel of biomarkers with clinical utility. Development of differential methylation diagnostic biomarkers could offer significant aid to patients requiring diagnosis and the health-care professionals struggling to provide diagnosis.