A Role for Genetic Modifiers in Tubulointerstitial Kidney Diseases

With the increased availability of genomic sequencing technologies, the molecular bases for kidney diseases such as nephronophthisis and mitochondrially inherited and autosomal-dominant tubulointerstitial kidney diseases (ADTKD) has become increasingly apparent. These tubulointerstitial kidney diseases (TKD) are monogenic diseases of the tubulointerstitium and result in interstitial fibrosis and tubular atrophy (IF/TA). However, monogenic inheritance alone does not adequately explain the highly variable onset of kidney failure and extra-renal manifestations. Phenotypes vary considerably between individuals harbouring the same pathogenic variant in the same putative monogenic gene, even within families sharing common environmental factors. While the extreme end of the disease spectrum may have dramatic syndromic manifestations typically diagnosed in childhood, many patients present a more subtle phenotype with little to differentiate them from many other common forms of non-proteinuric chronic kidney disease (CKD). This review summarises the expanding repertoire of genes underpinning TKD and their known phenotypic manifestations. Furthermore, we collate the growing evidence for a role of modifier genes and discuss the extent to which these data bridge the historical gap between apparently rare monogenic TKD and polygenic non-proteinuric CKD (excluding polycystic kidney disease).


Introduction
Tubulointerstitial kidney diseases (TKD) primarily involve the renal interstitium and tubular compartments.This typically results in tubulointerstitial fibrosis and tubular atrophy (IF/TA).The commonest TKDs include nephronophthisis and mitochondrially inherited, and autosomal-dominant tubulointerstitial kidney diseases (ADTKD).TKD results from a growing number of single-gene (monogenic) disorders.These are clinically characterised by a progressive decline in kidney function, leading to chronic kidney disease (CKD) and end-stage kidney disease (ESKD).TKD typically results in non-proteinuric CKD and is clinically differentiated from glomerular diseases by a lack of glomerular proteinuria and haematuria.However, more significant proteinuria can occur in the late stages due to secondary glomerulosclerosis.There may be impaired urinary concentrating ability and sodium reabsorption, resulting in polyuria and polydipsia.Extra-renal manifestations are highly variable (Table 1).In some instances, there may be renal cysts or structural abnormalities overlapping with polycystic kidney diseases (PKD) and congenital anomalies of the kidney and urinary tract (CAKUT), which are outside the scope of this review (Figure 1).

Karyomegalic Interstitial Nephritis (KIN)
Variably progressive CKD; IF/TA with enlarged and atypical tubular epithelial cell nuclei.

Renal tubular dysgenesis
Foetal anuria and perinatal death from pulmonary hypoplasia and oligohydramnios (Potter syndrome).
REN, AGT, AGTR1, ACE.Initially "lumped" under the term "nephronophthisis-medullary cystic kidney disease" complex, the availability of genomic sequencing technologies has led to more detailed phenotypic and molecular characterisation.However, genetic and phenotypic nomenclature are now used interchangeably, creating the potential for confusion within multidisciplinary teams comprising geneticists and nephrologists.As patients have received genetic diagnoses, it has become clear that older terms such as familial juvenile hyperuricaemic nephropathy (FJHN) and medullary cystic kidney disease (MCKD) have needed updating (Figure 1 and Table 1).Multiple investigators have since demonstrated that neither tubular microcysts nor larger cysts are pathognomonic for these diseases, nor does the medulla appear to be the specific location for cysts when occasionally observed [1][2][3][4].Some cases may be associated with hyperuricaemia and gout, but this is not pathognomonic and may be clinically silent in early disease [2,5].In an attempt to rectify this, the term ADTKD has been suggested by international guidelines, with the term ADTKD-NOS where the causal gene is unknown [6].Renal cysts and diabetes (RCAD) describes one of the potential manifestations of HNF1B variants that includes isolated TKD and congenital anomalies of the kidney and urinary tract (CAKUT).Nephronophthisis is characterised by IF/TA, tubular basement membrane abnormalities, and cystic dilatation of the collecting duct [7].Extra-renal manifestations, including several eponymous syndromes (Table 1), are seen in approximately one-fifth of nephronophthisis cases and are collectively known as NPHP-related ciliopathies (NPHP-RC) [8][9][10].A subset of ciliopathies recently named hepatorenal fibrocystic disease is characterised not only by renal IF/TA (with or without cysts) but also fibrosis and cystic dysgenesis of the liver and porto-biliary tract [11].Hepatorenal fibrocystic disease also includes autosomalrecessive polycystic kidney disease (ARPKD), characterised by congenital hepatic fibrosis and cystic dilatation of the renal collecting duct and most commonly resulting from variants in the ciliary IPT domain-containing fibrocystin/polyductin (PKHD1) [12].Autosomaldominant polycystic kidney disease (ADPKD) caused primarily by variants in polycystins 1 (PKD1) and 2 (PKD2) is also a ciliopathy; however, along with ARPKD, it falls outside the scope of this review, as cysts are the principal feature.Cystic dysplastic kidneys may occur alongside tubulointerstitial fibrosis in the lethal multisystem Meckel-Gruber syndrome.Cystic dysplasia or IF/TA may be present in Bardet-Biedl syndrome, short-rib thoracic dysplasia, and COACH syndrome (Table 1).

Mitochondrial Inherited Tubulointerstitial Kidney Disease (MITKD)
Mitochondrial cytopathies result from inherited or sporadic variants in mitochondrial DNA (mtDNA) or nuclear DNA that affect mitochondrial function and may present with nephrotic syndrome, Fanconi syndrome, TKD alone, or multisystem disorders [13][14][15].It has been suggested that the term mitochondrially inherited tubulointerstitial kidney disease (MITKD) is used to complement ADTKD [14].
Most autosomal-recessive TKD results from loss-of-function variants in genes encoding non-motile (primary) ciliary proteins.Cilia are organelles projecting from most mammalian cell types that communicate signals from the extracellular environment and other cells.Ciliopathies can result in isolated TKD, known as nephronophthisis (from the Greek for wasting of the nephron), as well as multisystem diseases (Figure 1 and Table 1).Over thirty genes are associated with nephronophthisis (Table 2); nephrocystin-1 (NPHP1) accounts for 20% of cases and other genes < 3% each [16].ADTKD is caused by heterozygous variants in any one of at least five genes, including UMOD, MUC1, HNF1B, REN, and SEC61A1 (Table 2).The best-studied is the UMOD gene encoding uromodulin, the most abundant protein in human urine [17].Recently, a phenotype with features of both TKD and PKD was attributed to variants in DNAJB11 [18].Mitochondrially inherited TKD is also well described [13].
Genetic TKD lies along a spectrum ranging from dramatic syndromic manifestations, typically diagnosed in childhood, to a more subtle phenotype difficult to distinguish from common forms of non-proteinuric CKD.Divergent phenotypes may be explained by variants of differing impact to gene functionality or mutated protein abundance.For example, homozygous variants in TMEM231, ranging from missense to null, are associated with diverse phenotypes, including Meckel-Gruber syndrome (MKS), orofaciodigital syndrome (OFD) type 3, and Joubert syndrome (JBTS) (Table 1) [19][20][21].Furthermore, variants in constrained gene regions are more likely to impact protein function exemplified by heterozygous variants in the EGF2, EGF3, and D8C domains of UMOD, resulting in earlier-onset ESKD [5,22].However, patients with the same primary pathogenic variant often manifest significant differences in penetrance or expressivity [23].The wide variation in age of onset of CKD and extra-renal manifestations, including distinct syndromic disorders, is not adequately explained by our understanding of single-gene disorders alone.This brings into question the historical dichotomy of kidney diseases as either monogenic or polygenic (Figure 2) [5,8,[24][25][26].Whilst tissue mosaicism has been demonstrated in ADPKD and Alports, there is no evidence for this in TKD, which suggests that mosaicism is not modifying variable expressivity [27].
Monogenic forms result from rare variants with large effect sizes often manifesting severe disease (but that may show variable penetrance), whereas polygenic forms result from the cumulative effect of multiple common variants, causing relatively milder disease.However, recent data suggest that monogenic disease risk may vary substantially due to polygenic background [28,29].Common variants in genes known to cause rare monogenic diseases are also associated with markers of CKD (e.g., UMOD, MUC1, IQCB1 (NPHP5), SDCCAG8, and IFT172) [30][31][32].This is further supported by gene-burden-association tests on 450,000 UKBioBank whole exomes, available at genebass.org,showing that UMOD, PKD2, and SLC22A2 are most statistically associated with CKD [33].
The concept of genetic modifiers was first introduced in 1941 by Haldane, but multiple definitions have since been suggested [34].Herein, modifier genes are defined as genes that alter the disease phenotype but are not required for the primary disease to be present.This is distinct from digenic or oligogenic inheritance, whereby two or more genes are essential for the manifestation of the primary disease phenotype [35].Genetic modifiers can affect the expression of another gene at multiple different organisational levels, including transcription, protein interactions, or the cellular and organ level [36].Terms such as epistasis, digenic/oligogenic inheritance, or modifier genes are often used interchangeably to describe the effect of one gene/allele on the phenotypic outcome of another gene/allele [35].For tri-allelic inheritance (which can be synonymous with digenic or oligogenic inheritance), at least three mutated alleles are required to manifest a primary disease phenotype.
Most typically, this involves both alleles of one gene and at least one allele in a second gene; however, all three variants can occur on the same gene (Figure 2) [37].Modifiers can be additive or suppressive and can affect penetrance, expressivity, and dominance.This review collates the evidence for the role of modifier genes and oligogenic inheritance on the phenotypic spectrum of TKD and attempts to bridge the gap between apparently rare monogenic TKD and polygenic non-proteinuric CKD.Whilst epigenetic, environmental, and non-coding factors may also play a role, these are outside the scope of this review.Simplified schematic representation of complex genetic inheritance from monogenic to polygenic (boxes represent genes; red vertical lines represent the pathogenic allele).A spectrum of increasingly complex diseases may occur with additional mutated alleles from monoallelic through to multiallelic.Additional complexity arises due to the number of genes affected from one gene (monogenic), a few genes (oligogenic), or many genes (polygenic).Modifiers are variants that can alter the disease phenotype but are not required for the primary disease to be present.Modifiers can be additive or suppressive and can affect penetrance, expressivity, and dominance.

The Evidence for Modifier Genes
Modifier effects may be collectively common but are likely to be individually rare and heterogeneous and have therefore largely eluded discovery in underpowered studies.
Most evidence comes from candidate gene or pathway approaches, with the more common variants likely to be tested first.Evidence for rare modifiers comes primarily from family case studies and are shown in Table 2.

NPHP1
In 25% of isolated nephronophthisis cases, a large genetic deletion arises from homologous recombination of genetic repeats, resulting in deletion of 290 kb and the entire NPHP1 gene (83 kb) [64,101,102].Wide variation in age of ESKD (pre-puberty to seventh decade) amongst patients homozygous for this deletion suggests modifier genes contribute to the variable phenotype [41,103].Hoefele et al. identified siblings with homozygous deletions of NPHP1, whereby both had nephronophthisis and retinitis pigmentosa.However, one sibling also carried a rare heterozygous variant in NPHP4 and developed ESKD aged nine (eight years earlier than her sibling) [40].
In a cohort of patients with nephronophthisis and neurological phenotypes consistent with Joubert syndrome (JBTS), half of those with variants in NPHP1 also harboured a second variant, including heterozygous missense variants in AHI1 and truncating variants in CEP290 [41].In another study, the same hypomorphic variant in AHI1 was associated with retinal disease in 153 patients with nephronophthisis irrespective of the underlying genetic cause of nephronophthisis [74].Functional evidence for this modifier is seen in homozygous Nphp1 knock-out mice (Nphp−/−) crossed with heterozygous Ahi1 knockouts (Ahi1+/−) demonstrating more significant degenerative retinal lesions [74].

NPHP3
Autosomal-recessive variants in NPHP3 (Nephrocystin-3) were initially thought to cause adolescent-or adult-onset nephronophthisis [42].However, recent case reports describe much earlier onset of ESKD, with some evidence for the role of genetic modifiers [107].In addition to pathogenic compound heterozygous variants in NPHP3, additional heterozy-gous frameshift or missense variants in NPHP4 are described in a number of families associated with earlier disease, including antenatal mortality [40,43].

CEP290
Autosomal-recessive variants in CEP290 (NPHP6) are implicated in an extensive range of syndromic clinical ciliopathies, including LCA, SLS, JBTS, BBS, and MKS (Figure 3) and also isolated nephronophthisis [45,46,48,[108][109][110]. Variant type and location do not adequately explain the highly variable phenotype [111].A deep intronic CEP290 variant (c.2991 + 1655A > G) resulting in aberrant splicing was identified in 16 (21%) of 76 unrelated patients with Leber congenital amaurosis (LCA).Despite altered splicing, a small amount of normally spliced protein was still present.This may explain the normal cerebellar and renal function in patients with LCA secondary to CEP290 variants in a dosage-dependent mechanism, with complete loss of CEP290 protein function in more severe manifestations like JBTS [108].However, differential phenotypes could equally be influenced by tissuespecific variation in gene isoform expression due to alternative splicing [112,113].This has been previously observed for WNK1, where variants are alternatively expressed on different tissue-specific isoforms in the kidney and nervous system, resulting in either hyperkalaemic hypertension or hereditary sensory and autonomic neuropathy type 2, respectively [114].Dose-dependent disease variants including alternative splicing and tissue-specific isoform expression in CEP290 are unlikely to fully account for the observed phenotypic heterogeneity.A study screening patients with the same CEP290 genotype but varied neurological severity revealed a novel heterozygous missense variant in AHI1 in a more severely affected patient [47].Furthermore, barttin CLCNK-type accessory subunit beta (BSND), historically associated with Bartter syndrome type 4a, has been identified to modify the severity of cystic kidney disease and renal failure progression in JBTS caused by CEP290 variants [49].Functional studies reveal a synergistic effect on the severity of phenotype with CEP290 and knockout of IQCB1 or CC2D2A [75,115].Other cilia components have also been shown to genetically or physically interact to modulate CEP290 phenotypes in mouse models [116].

TTC21B
Homozygous variants in tetratricopeptide repeat domain 21b (TTC21B) are associated with an array of different clinical entities including nephronophthisis, nephrotic range proteinuria, focal segmental glomerulosclerosis (FSGS), or global sclerosis [52].Amongst a cohort of patients with a "ciliopathy", pathogenic alleles in TTC21B were identified in 38/753 (5%) patients, suggesting TTC21B variants commonly contribute to the overall variant burden.Seventeen patients carried heterozygous variants in at least one of 13 other genes (including NPHP4, RPGRIP1L, TMEM216, CC2D2A, and MKS1) implicated in nephronophthisis or associated syndromes, suggesting modified genetic activity [51].

UMOD
In addition to modifiers in other genes (in trans), a phenotype may be altered by additional variation occurring in the same gene (in cis) and on the same haplotype/chromosome as the primary disease variant (Figure 2), with exemplars seen in cystic fibrosis [123].UMOD variants contribute to both monogenic TKD and polygenic forms of CKD.Common risk variants for the development of CKD and hypertension in the promoter region of UMOD are thought to have risen in population frequency due to selective pressure from increased urinary uromodulin defending against urinary tract infections [31,[82][83][84][85][86].A very high percentage of individuals have at least one risk variant that increases the quantity of uromodulin expressed (70 to 95%) [31,86].These variants are highly likely to co-occur in monogenic ADTKD-UMOD and may explain some of the extreme variation of progression to ESKD within families with the same pathogenic variant.A study of 147 families with monogenic ADTKD-UMOD identified underrepresentation of a protective allele (rs4293393-associated with reduced uromodulin production) compared with large population databases.This protective allele was linked to monogenic UMOD variants in only 11.6% of affected families but was in cis with "wild-type" UMOD in 17%.This compares with an expected minor allele frequency of 18-20% from the Genome Aggregation Database (gnomAD) [26].The authors postulate that this decreased protective allele frequency may be due to decreased expression of mutated uromodulin being less likely to receive a molecular diagnosis due to a milder clinical phenotype (with later development of ESKD) [73].Rare cases of homozygous UMOD variants in two consanguineous families demonstrate a more severe phenotype than with heterozygous variants [89,124].

MUC1
ADTKD-MUC1 frequently results from the insertion of an additional cytosine into a variable number of tandem repeats (VNTRs), resulting in a frameshift variant in MUC1 [90,122].The high-guanosine/cytosine (GC) content of the VNTR region hinders short read sequencing, and therefore, ADTKD-MUC1 is probably underdiagnosed.The age of onset of ESKD ranges from 16 to 80 years [1].With an allele frequency of 42% (gnomAD genomes), a common variant in MUC1 (rs4072037) is likely to coexist with monogenic forms of ADTKD-MUC1.It is an attractive genetic modifier candidate, as it influences gene expression through alternative splice-site mechanisms and is associated with declining kidney function in GWAS [30].

HNF1B
Heterozygous variants in HNF1B or 17q12 microdeletions encompassing the HNF1B gene are associated with large intra-familial variation in TKD (with and without cysts), maturity-onset diabetes of the young type 5 (MODY5), CAKUT, and other organ in-volvement, including neuropsychiatric symptoms [125].Structural kidney abnormalities have been significantly associated with splice-site variants and MODY5, specifically with frameshift variants [126].However, an extensive retrospective analysis of 377 patients with HNF1B kidney disease revealed no correlation between the genetic variant and renal failure severity [127].Although the co-occurrence of neurodevelopmental disorders may be due to other genes in the microdeletion, more recent data suggest a role for epigenetic modifiers due to differential methylation patterns [128,129].Evidence for the role of genetic modifiers is emerging, with heterozygous HNF1B variants detected with heterozygous PKD1 variants in a patient with early and more severe polycystic kidney disease [130].Another possible explanation for the wide phenotypic variability could be the influence of HNF1B as a transcription factor on the transcription of multiple other genes and, therefore, the specific variant burden in each of these downstream genes (Figure 4).

Mitochondrial Function
There are hundreds of mitochondrial DNA copies in each cell, and therefore, mutated copies may exist with normal copies in a state known as heteroplasmy.Disease expression depends on the proportion of dysfunctional mitochondria and tissue distribution [131].Multiple organs may be affected, including the kidneys, skeletal muscle, and central nervous system.The range of kidney diseases associated with a m.3243A > G variant in the tRNA Leu gene includes TKD and isolated tubulopathies, cystic kidney diseases, FSGS, as well as syndromic forms, including MELAS (myopathy, encephalopathy, lactic acidosis, and stroke-like episodes) and maternally inherited diabetes and deafness (MIDD) syndrome [98,132].One family with MIDD syndrome also carried a second variant in tRNA lys , which was absent in 75 controls.Isolated TKD phenotypes also occur in cases where all copies of (homoplasmic) mitochondrial DNA are mutated [13].
Secondary mitochondrial dysfunction also occurs in ADTKD-UMOD, resulting from the unfolded protein response and increased ER stress, leading to a decrease in the number of mitochondria and associated proteins [133].Any additional modifiers in the mitochondrial genome could affect the phenotype further (Figure 4).

Evidence for Oligogenic Inheritance
Bardet-Biedl syndrome (BBS) is a multisystem disorder underpinned mainly by dysfunction of primary cilia (Table 1 and Figure 4).Renal involvement in BBS is highly variable, including urine retention due to aberrant water trafficking, TKD, cystic dysplasia, hydronephrosis, and CAKUT.Despite wide renal phenotypic heterogeneity, no genotype correlation has been identified [134].A multi-protein complex called the BBSome comprises protein subunits encoded by BBS genes.BBS was initially thought to be a monogenic recessive disorder; however, deviation from the expected autosomal-recessive inheritance pattern in many pedigrees suggested alternative inheritance models.Following the screening of a large cohort of BBS families for variants in BBS2 and MKKS, tri-allelic inheritance was demonstrated in affected individuals from four pedigrees and revealed unaffected individuals from two pedigrees that carried two BBS2 variants but none in MKKS [135].This is an example of digenic, tri-allelic inheritance (Figure 2).Since then, tri-allelic inheritance has been identified in other cohorts of BBS patients and other diseases [37,[136][137][138].Beales et al. analysed 259 families and found evidence of tri-allelic inheritance and a homozygous missense variant of p.Met390Arg (the most frequent BBS1 variant in European populations) in asymptomatic members of two families, suggesting either incomplete penetrance or tri-allelic inheritance [136].Other studies have failed to identify tri-allelic inheritance in affected cases or have identified unaffected cases with bi-allelic inheritance [137][138][139][140][141][142][143] In other pedigrees, disease was identified in family members with rare double-homozygous BBS2 and BBS4 variants but not in tri-allelic first-degree relatives [48,144].
In a recent systematic secondary-variant burden analysis of patients with known bi-allelic variants in 1 of 17 BBS genes, researchers observed, in two independent cohorts, a non-random twofold enrichment of ultra-rare variants in other BBS genes compared with other recessive alleles and population controls [145].The suppression of 19 gene pairs in zebrafish revealed additive or suppressive effects and significant over-representation of secondary variants in BBS complex chaperonin genes.
Although most reports of BBS involve rare variants, functional evidence exists of a common modifier (1.4% of controls) in CCDC28B to disease severity [137].Given its population frequency, it may have been missed by traditional variant-filtration approaches.
The relatively high incidence of renal developmental abnormalities and renal cell carcinoma in BBS patients' relatives may represent heterozygous carriers [146].These findings suggest that tri-allelic inheritance results from modifier genes' action rather than oligogenic inheritance per se and that the degree of penetrance or expressivity is related to the cumulative effect of the variants.Gene regulatory networks in tubulointerstitial kidney disease with evidence of modifier genes, tri-allelic inheritance, or gene interaction.(a) Non-motile (primary) cilia dysfunction is a recognised driver of kidney disease, including nephronophthisis, polycystic kidney disease, or cystic dysplasia (Figure 1).Cilia are membrane-bound organelles that resemble finger-like projections on the apical surface of many tissues, including renal tubular epithelia.The primary cilium is composed of a basal body consisting of triplets of microtubules (grey) from which the cilium assembles [147].(b) Hotspots for ciliopathies include the intraflagellar transport proteins (green) responsible for protein trafficking and signalling pathways, the inversin compartment (blue) controlling cell polarity, the transition zone (purple) involved in the cell cycle and control of the entry and exit of proteins, and the BBSome (Bardet-Biedl syndrome proteins) (red) responsible for the trafficking of proteins to the cilia [147].This network of selected genes with evidence of modifier and tri-allelic inheritance reveals the close functional interaction of genes encoding the primary cilia.Solid lines represent modifier genes, and dashed lines represent genes with evidence for possible tri-allelic inheritance.The direction of the arrow denotes the direction of the contribution of additional variants to the primary mutated gene (e.g., homozygous variants in NPHP1 and a third mutated allele in AHI1 may result in a neurological phenotype like Joubert syndrome) [41].Bidirectional arrows indicate the equal contribution of variants.(c) There is significant potential for cumulative variantal burden in genes downstream of the transcription factor hepatocyte nuclear factor 1β (HNF1B) to modify disease phenotypes.Evidence in mouse models suggests that several genes are downregulated with Hnf1b inactivation (dashed line) [92].Downregulation of UMOD may also explain why most patients with HNF1B variants have hyperuricaemia typical of ADTKD-UMOD [148].Loss of HNF1B results in activation of a transcriptional network that induces extracellular matrix deposition and aberrant transforming growth factor 1 beta (TGF1B) signalling, resulting in tubulointerstitial fibrosis [149].HNF1B is linked to mitochondrial dysfunction in renal epithelial cells in experimental studies through other transcription factors [150].Mitochondrial dysfunction in ADTKD-UMOD also results secondary to endoplasmic reticulum stress from retention of misfolded uromodulin protein [133].Direct evidence of genetic modification has been reported in a patient with heterozygous variants in both PKD1 and HNF1B, causing early-onset severe disease [130].

Discussion
The historical dichotomy of TKD as either monogenic or polygenic is overly simplistic.Evidence strongly suggests that TKD can be impacted by a diverse set of modifiers as opposed to the simpler traditional model of one-gene, one-phenotype [37,[135][136][137][138][139][140][141][142][143][144]146,151,152].The high prevalence of disease-specific modifier variants suggests their broad role in TKD and even CKD of all causes [32,44,137].Common risk variants may prove significant when present in the same monogenic disease gene, for example, when in cis with primary pathogenic variants of UMOD and MUC1 [30,31].Monogenic disease suffers from the "Winner's curse", leading to a curtailing of genetic investigation once a single variant is identified.However, penetrance and expressivity of TKD are likely to be governed by the total variant burden both within individual genes and across multiple gene pathways involved in the structure and function of the tubulointerstitium.The pleiotropic effects of pathogenic variants in HNF1B demonstrate the potential impact of gene networks [119,153].Multiple genetic "hits" in these TKD genes are likely to contribute variably to the disease spectrum, from apparent monogenic to polygenic or multifactorial disease patterns.Here, we describe severe rare modifiers and common moderate modifiers.However, it is essential to consider the role of multiple smalleffect variants traditionally associated with polygenic diseases whose overall burden may be cumulatively large in some individuals yet seldom identified due to weak association.Most studies focus on modifiers that result in more severe diseases; however, patients with milder phenotypes who are less likely to receive a molecular diagnosis are equally likely to be subject to modifier genes.Only a minority of patients with TKD have overt extra-renal or syndromic manifestations, and even the renal phenotype may be subtle.With bland urinalysis or occasionally mild proteinuria and an association with renal cysts that is neither universal nor pathognomonic, TKD is difficult to differentiate from other non-proteinuric CKD [6,154,155].
Several barriers remain to understanding the more complex genetics of TKD, including poor differentiation of TKD from other causes of CKD, inadequate phenotyping and nomenclature, and methodological challenges, and biases in genetic analyses.
Recent evidence has identified relatively "mild", adult-onset disease with recessive inheritance and, conversely, severe paediatric-onset disease resulting from dominant gene variants.For example, autosomal-recessive nephronophthisis due to NPHP1, the commonest genetic cause of paediatric ESKD, has recently been implicated in at least 0.5% of adult-onset ESKD [9,16,103].Conversely, autosomal-dominant TKD resulting from variants in HNF1B and REN can cause severe disease in childhood [127,156].Molecular genetics has exposed the limitations of traditional clinical diagnostics not only due to incomplete phenotyping but with phenocopies increasingly identified following molecular genetic testing [157,158].Attempts to diagnoses have led to multiple and sometimes misleading nomenclature that in some cases has been used to predefine limited molecular diagnoses, such as the use of ADTKD-NOS (not otherwise specified).
Multiple and sometimes misleading disease names predefine limited molecular diagnostics and reinforce bias (Figure 1).Attempts to increase diagnoses have led to terms such as autosomal-dominant tubulointerstitial kidney disease (ADTKD) even before a molecular diagnosis has been made [6].
Although massive datasets of normal variant distribution are now available, these allow comparison of the burden of rare variants on a population level but not on an individual level and often fail to provide contextual information for multiple variants.Additionally, extensive genomic datasets have historically been biased towards white European ancestry.
GWAS data implicate coding and non-coding variants with a variety of kidney traits, and these may prove important in the polygenic inheritance patterns of TKD.The challenge facing the research community is in the use of these data to identify variants that may contribute to the risk of disease progression or the emergence of TKD.One approach may be utilising resources that map expression quantitative trait loci (eQTLs) across multiple human tissues, aiding in the prioritisation of disease-causing genes [159].
Genetic modifiers are methodologically challenging to detect and expensive to prove functionally.Classifying variants as benign or pathogenic requires multiple types of evidence [160].The challenges to functional modelling of genetic variants in model systems for single-variant disorders scale exponentially with additional component alleles.The more alleles required to be modelled, the greater the experimental design's time, cost, and complexity.
Improved definition of the molecular mechanisms underlying TKD is required to inform prognosis and shift the emphasis from reactive therapies to secondary or primary prevention.Extensive prospective NGS analyses have the potential to unravel the genetic and phenotypic complexities of so-called monogenic kidney diseases, not least of all TKD.

•
Tubulointerstitial kidney disease may present with a severe syndromic phenotype traditionally diagnosed in childhood, yet most TKD patients manifest a subtle phenotype with little to differentiate them from other common forms of non-proteinuric chronic kidney disease.

•
Phenotypes may vary even within families that share the same putative monogenic gene variant and shared environmental factors.

•
Common variants in monogenic TKD genes are also associated with chronic kidney disease at the population level.

•
There is growing evidence for tri-allelic inheritance as well as for rare modifiers of severe effect and common modifiers of moderate effect on patient phenotypes.

Figure 2 .
Figure 2.Simplified schematic representation of complex genetic inheritance from monogenic to polygenic (boxes represent genes; red vertical lines represent the pathogenic allele).A spectrum of increasingly complex diseases may occur with additional mutated alleles from monoallelic through to multiallelic.Additional complexity arises due to the number of genes affected from one gene (monogenic), a few genes (oligogenic), or many genes (polygenic).Modifiers are variants that can alter the disease phenotype but are not required for the primary disease to be present.Modifiers can be additive or suppressive and can affect penetrance, expressivity, and dominance.

Figure 3 .
Figure 3.The spectrum of disease associated with variants in centrosomal protein 290 (CEP290).(a) Eye involvement ranges from a relatively mild retinal phenotype to Leber congenital amaurosis (LCA), which is a leading cause of childhood blindness.(b) Senior-Løken syndrome is characterised by retinal dystrophy and nephronophthisis.(c) Joubert syndrome (JBTS) adds cerebellar vermis hypoplasia to retinal dystrophy and nephronophthisis.(d) Bardet-Biedl syndrome is truly multisystem, with significant features of retinal disease, central obesity, postaxial polydactyly, cognitive impairment, genitourinary abnormalities, and kidney disease.(e) Meckel-Gruber syndrome is a lethal multisystem disorder with occipital meningo-encephalocele, hepatobiliary ductal plate malformation, postaxial polydactyly, and cystic dysplasia of the kidneys with marked interstitial fibrosis.

Figure 4 .
Figure 4. Gene regulatory networks in tubulointerstitial kidney disease with evidence of modifier genes, tri-allelic inheritance, or gene interaction.(a) Non-motile (primary) cilia dysfunction is a recognised

Table 2 .
Monogenic causes of TKD and evidence of modifier genes.