The genetic landscape of autosomal dominant polycystic kidney disease in Kuwait

ABSTRACT

Background

Autosomal dominant polycystic kidney disease (ADPKD) is the most common renal monogenic disease, characterized by bilateral accumulation of renal fluid-filled cysts leading to progressive renal volume enlargement and gradual impairment of kidney function, often resulting in end-stage renal disease. Kuwait could provide valuable genetic insights about ADPKD, including intrafamilial phenotypic variation, given its large household size. This study aims to provide a comprehensive description of the pathogenic variants linked to ADPKD in the Kuwaiti population using multiple genetic analysis modalities and to describe and analyse the ADPKD phenotypic spectrum in terms of kidney function, kidney volume and renal survival.

Methods

A total of 126 ADPKD patients from 11 multiplex families and 25 singletons were recruited into the study. A combination of targeted next-generation sequencing (tNGS), long-range polymerase chain reaction, Sanger sequencing and multiplex ligation-dependent probe amplification were utilized for genetic diagnosis. Clinical evaluation was conducted through renal function testing and ultrasonographic kidney volume analysis.

Results

We identified 29 ADPKD pathogenic mutations from 36 families achieving an overall molecular genetic diagnostic rate of 112/126 (88.9%), including 29/36 (80.6%) in families. A total of 28/36 (77.8%) families had pathogenic mutations in PKD1, of which 17/28 (60.7%) were truncating, and 1/36 (2.8%) had a pathogenic variant in the IFT140 gene. A total of 20/29 (69%) of the identified ADPKD mutations were novel and described for the first time, including a TSC2-PKD1 contiguous syndrome. Clinical analysis indicated that genetically unresolved ADPKD cases had no apparent association between kidney volume and age.

Conclusion

We describe for the first time the genetic landscape of ADPKD in Kuwait. The observed genetic heterogeneity underlining ADPKD along with the wide phenotypic spectrum reveal the level of complexity in disease pathophysiology. ADPKD genetic testing could improve the care of patients through improved disease prognostication, guided treatment and genetic counselling. However, to fulfil the potential of genetic testing, it is important to overcome the hurdle of genetically unresolved ADPKD cases.

Graphical Abstract

Graphical Abstract.

INTRODUCTION

Autosomal dominant polycystic kidney disease (ADPKD) is the most common renal monogenic disease, affecting ≈1/1000 individuals, and is the fourth leading cause of end-stage renal disease (ESRD) in adults [1, 2]. It is a systemic disease characterized by bilateral accumulation of renal fluid-filled cysts, leading to progressive renal volume enlargement and gradual impairment of kidney function, often resulting in ESRD [3]. ADPKD patients can also develop extrarenal manifestations such as polycystic liver disease (PLD), cerebral aneurysms and cardiovascular abnormalities, including early-onset hypertension [4]. Around 78% of ADPKD cases are linked to pathogenic variants in PKD1 and ≈15% to pathogenic variants in PKD2, while 7% of ADPKD cases remain genetically unresolved or are due to pathogenic variants in other genes [5, 6]. PKD1 and PKD2 encode polycystin-1 (PC1) and polycycstin-2 (PC2), respectively, which are membrane glycoproteins forming a 1:3 heteromeric complex and function, most likely, in the primary cilium [7–9]. ADPKD severity shows some degree of complexity, where mutations in PKD1 have shown a more severe disease phenotype than mutations in PKD2 in terms of total kidney volume and onset of ESRD (54.3 years for PKD1 and 79.7 years for PKD2) [10–14]. Furthermore, truncating mutations in PKD1 are associated with a shorter time to ESRD when compared with PKD1 non-truncating mutations (55.6 and 67.9 years, respectively) [15].

Recent advances in genetic sequencing technologies have allowed wider utilization of next-generation sequencing (NGS) platforms, including whole genome sequencing (WGS), whole exome sequencing (WES) and targeted sequencing (tNGS), to explore the genetic basis of the heterogeneous spectrum of ADPKD-like phenotypes that is unrelated to mutations in either of PKD1 or PKD2. A recent report has shown that IFT140 is the third most common ADPKD spectrum gene after PKD1 and PKD2. Monoallelic pathogenic variants in this gene, which encodes a component of the intraflagellar transport (IFT) complex that is involved in the formation and maintenance of cilia, have been associated with a milder spectrum of the ADPKD phenotype with limited renal insufficiency [16]. Other genes causative of rarer forms of the disease include GANAB, DNAJB11 and ALG9. Monoallelic mutations in GANAB, which encodes the glucosidase IIα subunit, have been associated with a milder form of ADPKD and autosomal dominant polycystic liver disease (ADPLD) [17]. Monoallelic mutations in DNAJB11, which encodes a soluble glycoprotein that is involved in protein processing in the endoplasmic reticulum, have been shown to cause an atypical presentation of ADPKD with phenotypic overlap with a milder form of ADPKD and autosomal dominant tubulointerstitial diseases (ADTKDs) characterized by cystic kidneys with normal size and progressive interstitial fibrosis [18]. Other reports have shown that mutations in ALG9, which encodes alpha-1,2-mannosyltransferase enzyme, are associated with a milder form of ADPKD [19].

The observed genetic heterogeneity underlining ADPKD along with the wide phenotypic spectrum reveal the level of complexity in disease pathophysiology. A recent ‘threshold model’ indicated that renal cyst formation is likely to be triggered by the reduction in functional levels of polycystin in tubular epithelial cells beyond a critical threshold. This could be a result of either mutations in PKD1 and/or PKD2 or mutations in other genes, including GANAB, DNAJB11 and ALG9, that affect the endoplasmic reticulum protein biosynthesis, which in turn have a negative functional impact on the polycystins [20].

Currently ADPKD genetic testing is clinically indicated in limited scenarios, including to confirm or exclude an ADPKD diagnosis in suspected cases with no apparent family history or with equivocal imaging findings. It is also used to confirm or exclude an ADPKD diagnosis in young patients for kidney donation decisions. As it is the most common inherited kidney disease, genetic testing for ADPKD is a rapidly increasing area [21]. Applying ADPKD genetic testing in the clinical context could improve the care of patients through improved disease prognostication, treatment guidance, support of clinical trials and as an aid in genetic counselling and pre-implementation genetic diagnosis applications [21–23].

The genetic landscape of ADPKD in Kuwait has not been explored previously. As a small country with an average household size of seven people [24], Kuwait could provide valuable genetic insights about ADPKD, including intrafamilial phenotypic variation. In this study we provide a comprehensive description of the pathogenic variants linked to ADPKD in the Kuwaiti population using multiple genetic analysis modalities. We further describe and analyse the phenotypic spectrum of ADPKD in the studied population in terms of kidney function, kidney volume and renal survival.

MATERIALS AND METHODS

Patient recruitment

Participants were recruited from the outpatient nephrology clinic, inpatients and dialysis clinic and kidney transplant outpatients at Mubarak Al-Kabeer Hospital. Patients were recruited into the study if they satisfied the inclusion criteria. All patients are ≥18 years of age, able to provide informed consent and without previous genetic results related to ADPKD. Patients included had a positive family history of ADPKD and met the unified criteria for ultrasonographic diagnosis of ADPKD [25] or were without a family history of ADPKD but with confirmed renal cysts on imaging as per the unified criteria for ultrasonographic diagnosis of ADPKD [25].

The study was reviewed and approved the Ministry of Health (MOH) Research Ethics Committee (reference 1139/2019) and the Joint Committee for the Protection of Human Subjects in Research of the Health Sciences Center at Kuwait University and the Kuwait Institute for Medical Specialization (reference VDR/JC/690). Written informed consent was obtained from all enrolled patients in accordance with the guidelines of the MOH Research Ethics Committee.

Clinical evaluation

Enrolled patients were evaluated using ultrasonographic analysis and renal function tests to confirm their disease status. All enrolled patients, except for those with ESRD or kidney transplant, underwent abdominal sonographic examination using a Logic 7 ultrasound with a curvilinear 3.5-MHZ probe (GE Healthcare, Chicago, IL, USA). Additional imaging was performed by both non-enhanced and contrast-enhanced multidetector computed tomography with BrightSpeed computed tomography (CT) scanners (GE Healthcare) with multiplanar reformation. Magnetic resonance imaging (MRI) was performed as multisequence multiplanar non-enhanced using a 1.5-Tesla machine (GE Healthcare). The scan was performed to examine the kidneys, liver and pancreas, as well as the ovaries in female subjects. Each kidney was examined in multiple views. Renal cysts were examined and counted for diagnostic purposes according to the unified criteria for ultrasonographic diagnosis [25]. The total kidney volume (TKV) for each kidney was calculated using the ellipsoid formula: volume = length × lateral diameter × anterior–posterior diameter × π/6. The machine calculated the TKV automatically in cubic centimetres (cc) and then it was adjusted for height (htTKV, expressed as cc/m). The liver and pancreas of each patient were screened for the presence of cysts. For the renal function test (RFT), a 5-ml blood sample was taken from each enrolled patient, except patients with documented ESRD or kidney transplant. All RFTs were performed at the Mubarak Al-Kabeer Hospital, Jabriya, Kuwait. Estimated glomerular filtration rates (eGFRs) were calculated using the Chronic Kidney Disease Epidemiology Collaboration creatinine equation [26].

DNA isolation

A 10-ml blood sample from each enrolled patient was collected in ethylenediaminetetraacetic acid tubes. Genomic DNA was isolated from each blood sample using the Gentra Puregene Blood Kit (Qiagen, Hilden, Germany; catalogue no. 158467) following the manufacturer's protocol.

tNGS

DNA libraries were prepared for hybridization-based targeted capture of a 65-gene, 136-gene or 357-gene panel as previously described [6, 16, 18]. Captured libraries were sequenced on an HiSeq4000 (Illumina, San Diego, CA, USA) with 2 × 150 paired-end reads. Read alignment and variant calling were performed in the Genome Analysis Toolkit and variant mining completed in the Sequence and Variation Suite (Golden Helix, Bozeman, MT, USA) as previously described [16, 18]. Samples were evaluated for copy number variants utilizing the log ratio of observed read depth over expected read depth, based on normalized sample coverage in the target region [6, 16].

Long-range polymerase chain reaction (LR-PCR), Sanger sequencing and multiplex ligation-dependent probeamplification (MLPA)

Variants of interest (VOIs) identified on the panel were confirmed and segregated in additional family members by PCR for Sanger sequencing using established primer pairs and conditions [5]. For variants in the duplicated region of PKD1, LR-PCR was performed prior to exon-specific amplification using the previously described primer sets [6]. Regions of PKD1 and PKD2 that were poorly covered by the panel were further screened by LR-PCR methods in samples without a VOI identified by tNGS [16]. Large deletions in PKD1 or PKD2 were confirmed by established MLPA assays (MRC Holland, Amsterdam, The Netherlands) and evaluated in GeneMarker (SoftGenetics, State College, PA, USA). Additionally, a subset of samples was screened in full by the LR-PCR, Sanger and MLPA methodologies without prior panel analysis.

Variant pathogenicity and analysis

Variants were classified as VOIs if they were ultrarare (<10× for known dominant genes and <50× for recessive genes in gnomAD version 2.1.1) and predicted to alter either the protein sequence or splicing. VOIs were then evaluated for pathogenicity utilizing the previously defined criteria and classified into mutation strength groups [5].

Data analysis

Renal survival analysis was performed to study the relationship between PKD1 mutation type (truncating and non-truncating) and time to ESRD. Survival times were calculated as the time of onset of ESRD. Kaplan–Meier product-limit renal survival curves were constructed for patients with PKD1 truncating and non-truncating mutations using MATLAB and Statistics Toolbox, release 2012b (MathWorks, Natick, MA, USA). The mean survival time was calculated as the shortest survival time in which the estimated probability of renal survival was ≤0.5. The estimation is limited to the largest survival time if it is censored. The differences between both groups were evaluated for statistical significance using the logrank (Mantel–Cox) test ,where significance was established for P-values <.05.

Multivariable linear regression was conducted to assess associations between age, htTKV and eGFR while stratifying by PKD1 mutation status. Moreover, interaction terms between age and PKD1 mutation status were assessed. Sex-stratified analyses were conducted to evaluate the association between PKD1 mutation status and htTKV and eGFR. These models were followed by tests for interaction by including a product term (sex × PKD1 mutation) in the multivariable linear regression models. htTKV was modelled as log10 transformed variable, hence taking the antilog of the linear regression coefficients (β) yields an adjusted ratio of geometric means.

RESULTS

Study participants

A total of 126 ADPKD patients from 11 multiplex families and 25 singletons fulfilled the inclusion criteria and were enrolled in the study. The mean age of the cohort was 37.29 years (range 18–60) and 74/126 (58.7%) were males (Table 1). Of the recruited patients, 52/126 (41.3%) had liver cysts, confirmed by ultrasound, and 92/126 (73%) were diagnosed with hypertension (Table 1). Only 33/126 (26.2%) of the recruited patients were already at ESRD at the time of enrolment (Table 1).

Table 1:

Baseline characteristics of ADPKD patients based on their genetic diagnosis status.

Characteristics	All patients (N = 126)	Solved (n = 112)	Unsolved (n = 14)	PKD1 truncating mutation (n = 74)	PKD1 non-truncating mutation (n = 36)
Age (years), mean (range)	37.29 (18–60)	38.32 (18–60)	40 (26–56)	37.93 (18–60)	37.74 (18–58)
Male, n/N (%)	74/126 (58.7)	65/112 (58)	9/14 (64.3)	41/74 (55.4)	21/36 (58.3)
Known liver cysts, n/N (%)	52/126 (41.3)	47/112 (42)	5/14 (35.8)	31/74 (41.9)	15/36 (41.7%)
Known hypertension, n/N (%)	92/126 (73)	83/112 (74.1)	9/14 (64.3)	55/74 (74.3)	25/36 (69.4)
Documented family history, n/N (%)	112/126 (88.9)	101/112 (90.2)	11/14 (78.6)	69/74 (93.2)	30/36 (83.3)
Patients reaching ESRD, n/N (%)	33/126 (26.2)	33/112 (29.5)	0/14 (0)	20/74 (27)	13/36 (36.1)

ADPKD genetic diagnosis

DNA samples from the 126 PKD patients were analysed for PKD genetic diagnosis. From the enrolled families, 28/36 (77.8%) had a pathogenic mutation in PKD1 (NM_001009944) and 1/36 (2.8%) had a pathogenic variant in the IFT140 gene (NM_014714) resulting in a resolve rate of 29/36 (80.6%). The remaining 7/36 (19.4%) were unresolved genetically (Fig. 1A).

Figure 1:

Pathogenic variants linked to PKD in the cohort. (A) Genes linked to ADPKD in the presented cohort. (B) Details on the genetic modalities utilized for successful detection of ADPKD pathogenic variants. (C) An overview of types of PKD1 pathogenic variants.

Following the Mayo Clinic PKD pathogenicity guidelines, we report an overall molecular genetic diagnostic rate of 112/126 (88.9%) and 29/36 (80.6%) in families (Table 1). From the solved families, 19/29 (65.5%) were diagnosed using tNGS, 8/29 (27.6%) were diagnosed using LR-PCR and Sanger sequencing and 2/29 (6.9%) were diagnosed using MLPA (Fig. 1B).

Detected variants

A total of 29 pathogenic variants were detected in the resolved families and there were 9 variants of possible PKD relevance (Table 2). For the 28 detected PKD1 pathogenic variants, 7 (25%) were nonsense, 7 (25%) were missense, 6 (21.4%) were frameshift indel, 4 (14.3%) were large deletion, 2 (7.1%) were splicing variants and 2 (7.1%) were in-frame indel (Table 2 and Figs. 1C and 2). From a total of 29 ADPKD pathogenic variants detected, 20 (69%) were novel variants, while for the other possible relevant PKD variants, 4/9 (44.4%) were novel (Table 2). Conservation analysis of the detected variants is presented in (Supplementary Table 2).

Table 2:

Details of ADPKD pathologic and possibly modifying variants.

Pedigree	Gene	Ex/IVS	Protein variant	cDNA variant	Mutation type	Functional effect	Mayo Clinic designation	gnomAD1	ADPKD Variant Database	dbSNP ID
Pathogenic variants
1	PKD1	5′UTR-Ex14	p.Met1fs	c.(1-2kb)_(3295)del23.5kb	Large deletion	Truncating	DP		No
2	PKD1-TSC2	2–46 and 18–41	PKD1: p.Leu72fs* TSC2: p.Ala1614fs*	Chr16: (2.121Mb)_(2.170Mb)del48.3kb-PKD1: c.(216)_(12909*)del30.9kb TSC2: c.(1840)_(5352)del17.4kb	Large deletion	Truncating	DP		No
3	PKD1	5	p.Glu226*	c.676G>T	Nonsense	Truncating	DP		No
4	PKD1	15	p.Met1219fs	c.3656_6911del3256	Large deletion	Truncating	DP		No
5	PKD1	15	p.Arg1672fs	c.5014_5015delAG	Frameshift	Truncating	DP		Yes	rs1555455457
6	PKD1	15	p.Gln2058Pro	c.6173A>C	Missense	Non-truncating	LP	0	No
7	PKD1	15	p.Arg2163*	c.6487C>T	Nonsense	Truncating	DP		Yes
8	PKD1	15	p.Arg2220_Pro2224del5	c.6657_6671del15	InFrame D/I	Non-truncating	DP		Yes
9	PKD1	15	p.Gln2243*	c.6727C>T	Nonsense	Truncating	DP		Yes	rs1567191609
10	PKD1	15	p.Gly2278Arg	c.6832G>A	Missense	Non-truncating	LP	0	Yes	rs1555454145
11	PKD1	15	p.Trp2298Arg	c.6892T>G	Missense	Non-truncating	HLP	0	No
12	PKD1	IVS15	p.Arg2306fs	c.6916-9G>A	Splice	Non-truncating	LP	0	Yes
13	PKD1	17	p.Cys2373*	c.7119C>A	Nonsense	Truncating	DP		Yes
14	PKD1	17	p.Arg2404fs	c.7197_7209dup13	Frameshift	Truncating	DP		No
15	PKD1	23	p.Phe2806Ser	c.8417T>C	Missense	Non-truncating	LP	0	No
16	PKD1	IVS24	p.Gly2983fs	c.8948+2dupT	Splice	Non-truncating	LP		No
17	PKD1	25	p.Glu3020*	c.9058G>T	Nonsense	Truncating	DP		Yes
18	PKD1	25	p.Leu3010Gln	c.9029T>A	Missense	Non-truncating	LP	0	No	rs750501225
19	PKD1	26	p.Pro3069fs	c.9205_9224del20insGACA	Frameshift	Truncating	DP		No
20	PKD1	36	p.Ser3593Arg	c.10779C>G	Missense	Non-truncating	LP	0	No
21	PKD1	39	p.Tyr3734*	c.11202C>A	Nonsense	Truncating	DP		No	[54]
22	PKD1	40	p.Tyr3759fs	c.11274_11275delCT	Frameshift	Truncating	DP		No	rs1555446105
23	PKD1	40	p.Asp3780_Asp3782del3	c.11339_11347del9	InFrame D/I	Non-truncating	HLP		No
24	PKD1	41	p.Trp3842Arg	c.11524T>C	Missense	Non-truncating	HLP	0x	Yes	rs1057518959
25	PKD1	IVS42-3′UTR	p.Val3905fs	c.11712+55_12909974del2.6kb	Large deletion	Truncating	DP		No
26	PKD1	43	p.Gln3955fs	c.11863dupC	Frameshift	Truncating	DP		No	[43]
27	PKD1	44	p.Gln4005*	c.12013C>T	Nonsense	Truncating	DP		Yes	rs1567148587
28	PKD1	46	p.Phe4219fs	c.12627_12655dup29	Frameshift	Truncating	DP		No
29	IFT140	14	p.Gly509?	c.1525-1G>A	Splice	Truncating	DP		NA	[16]
Other possible relevant variants
22	PKD1	8	p.Thr558Met	c.1673C>T	Missense	Non-truncating	LN	8x	No	rs781572938
Unsolved#	PKD1	15	p.Arg1549Gln	4646G>A	Missense	Non-truncating	LN	2x	No	rs530555146
9	PKD1	15	p.His1769Tyr	c.5305C>T	Missense	Non-truncating	Mod	5x	No
Unsolved	PKD1	23	p.Thr2756Ile	c.8267C>T	Missense	Non-truncating	VUS	49x	No	rs141296093
22	PKD1	27	p.Arg3169Trp	c.9505C>T	Missense	Non-truncating	LN	8x	No	rs765180455
21	PKD1	36	p.Lys3607Met	c.10820A>T	Missense	Non-truncating	VUS	0x	No
4	PKD2	15	p.Met905Leu	c.2713A>C	Splice	Non-truncating	VUS	77x	2	rs573871626
Unsolved	IFT122	30	p.Leu1257_Pro1263del	c.3769_3789+5del26	Frameshift	Truncating	DP (rec)	0x	No
Unsolveda	COL4A1	30	p.Pro739Leu	c.2216C>T	Missense	Non-truncating	VUS	13x	No

^aCo-inherited by the same patient.

dbSNP: Single Nucleotide Polymorphism Database; FS del: frameshift deletion; FS dup: frameshift duplication; IF del: in-frame deletion; DP: definitely pathogenic; LP: likely pathogenic; VUS: variant of uncertain significance; LN: likely neutral; Rec: recessive; HLP: highly likely pathogenic; Mod: moderate; Ex: exon; IVS: intervening sequence.

Figure 2:

The distribution of detected PKD1 variants. Variants highlighted in bold are pathogenic variants and those unhighlighted are the non-pathogenic variants with possible ADPKD relevance.

A case with the TSC2-PKD1 contiguous syndrome

Among the novel variants detected was a large deletion covering parts of PKD1 and TSC2 from PKD1 exon 2 to TSC2 exon 18. The full genomic location of the deletion is Chr16:(2.121Mb)(2.170Mb)del48.3kb. The TSC2 portion of the annotation is c.(1840)_(5352)del17.4kb(p.Ala1614fs*) and the PKD1 portion of the annotation is c.(216)_(12909*)del30.9kb(p.Leu72fs*) (Table 2 and Fig. 2). This contiguous gene syndrome deletion was detected in a 25-year-old male patient with early renal insufficiency (eGFR 9.8 ml/min/1.73 m²). Radiologic analysis revealed renal angiomyolipoma, pulmonary lymphangioleiomyomatosis and bilateral subependymal calcified nodules (Fig. 3A–E). The patient also showed multiple skin lesions including fibrous cephalic plaques, hypopigmented macules, angiofibroma and shagreen patches on the lower back.

Figure 3:

Diagnostic imaging of two rare PKD cases. (A) A CT scan of a patient with TSC2/PKD1 contiguous gene syndrome shows markedly enlarged kidneys with innumerable variable-sized cortical and parenchymal cysts, some showing scattered tiny mural calcifications. (B) Renal masses of mixed soft tissue and fat attenuation, suggestive of angiomyolipoma, are noted with a large amount of ascites. (C) A CT scan shows bilateral basal lung bronchiectatic changes along with atelectatic bands. (D and E) Small left hepatic lobe cysts were noted. (F and G) Multiple calcified subependymal nodules were noted. (H and I) A CT scan of the IFT140 patient shows enlarged kidneys by the presence of multiple variable-sized cortical and to lesser extent parenchymal renal cysts. (H) The largest cysts are the lower polar exophyting ones. (I) Small mural calcifications are seen at the right-sided lower polar cyst.

A case of co-inheritance of a PKD2 variant of uncertain significance (VUS) with a pathogenic PKD1 variant

In a rare case, genetic analysis revealed a 42-year-old singleton male co-inherited a large PKD1 deletion (c.3656_6911del3256) along with a PKD2 VUS, which may alter splicing (c.2713A>C), that was reported 77 times in gnomAD. The patient showed a more severe form of the disease, reaching ESRD at 39 years of age (Table 2 and Supplementary Table 2 and Supplementary Fig. 1).

IFT140 case of PKD

One pathogenic variant was detected in a gene other than PKD1. A splicing pathogenic variant at a conserved site was detected in IFT140 c.1525-1G>A with a truncating effect (Table 2 and Supplementary Table 1). This variant was detected in a 44-year-old female with normal eGFR and htTKV of 876 ml/m. The patient showed eight cysts in each kidney and no cysts in the liver (Fig. 3F–H).

Genetically unsolved PKD cases with inherited variants of PKD relevance

We detected four variants with possible PKD relevance in three unsolved cases (Table 2). In a 47-year-old female patient, which was not resolved genetically, a novel frameshifting variant (c.3769_3789+5del26) was detected in IFT122 (NM_052985), resulting in a frameshift mutation with a possible monoallelic pathogenic effect (Table 2). The patient had cysts in both kidneys with an htTKV of 539.5 cc/m and an eGFR of 78 ml/min/1.73 m². The patient had no family history of PKD.

Genetic analysis also revealed co-inheritance of a missense variant (c.2216C>T, p.Pro739Leu) in COL4A1 (NM_001845) with uncertain significance with a missense non-truncating PKD1 variant (c.4646G>A, p.Arg1549Gln) (Table 2) in a 21-year-old female with normal kidney volume and function (Table 2). We also detected a PKD1 (c.8267C>T, p.Thr2756Ile) missense variant with uncertain significance in a 41-year-old male with enlarged kidneys (htTKV 2019.9 cc/m) and reduced kidney function (eGFR 42 ml/min/1.73 m²) (Table 2).

Clinical and renal survival analysis

We utilized htTKV and eGFR to determine the clinical course in patients with PKD1 truncating and non-truncating mutations and the unsolved cases. Figure 4 shows an association of age with eGFR and kidney volume (htTKV) according to PKD1 mutation status. Patients with PKD1 truncating (sex adjusted: β = −2.10, P < .001) and PKD1 non-truncating (sex adjusted: β = −1.95, P = .019) mutations demonstrated decreasing eGFR values as age increased (Fig. 4A). Moreover, there was no evidence for a difference in the association between age and eGFR according to PKD1 mutation type [P for interaction (age × PKD1 mutation) = .501]. Fig. 4B shows that htTKV increased as age increased among those with PKD1 truncating mutations (sex adjusted: β = 23.65, P < .001) and PKD1 non-truncating mutations (sex adjusted: β = 24.72, P = .034), with no indication for a difference in the magnitude of the association across mutation status [P for interaction (age × PKD1 mutation) = .817] (Supplementary Table 1).

Figure 4:

Relationship between eGFR and htTKV in the ADPKD cohort. (A) Plot of eGFR versus age for ADPKD patients with a truncating PKD1 mutation (red) and patients with a PKD1 non-truncating mutation (blue) shows a decreasing trend in both groups. (B) Plot of the htTKV in natural log scale versus age for ADPKD patients with a truncating PKD1 mutation (red) and patients with a PKD1 non-truncating mutation (blue) shows an increasing trend for both groups. (C) A plot shows a decreasing trend of eGFR versus age for the unsolved cases and PKD1 solved cases. (D) Plot of htTKV versus age shows a significant difference in the htTKV trend for PKD1 solved cases. Shaded areas indicate 95% confidence intervals.

Similarly, when testing the association between eGFR and age according to genetic status (solved and unsolved cases) (Fig. 4C), there was evidence for a decreasing trend in eGFR as age increased in those resolved PKD1 cases (sex adjusted: β = −1.99, P < .001) and in those unresolved cases (sex adjusted: β = −1.83, P = .014). There was no evidence for heterogeneity in these associations by PKD1 mutation status [P for interaction (age × PKD1 mutation) = .925]. Moreover, kidney volume showed an increasing trend as age increased among PKD1-resolved cases (sex adjusted: β = 24.22, P < .001), whereas there was no association among unresolved cases (sex adjusted: β = −9.16, P = .643; Fig. 4D). There was statistical evidence that the association between age and htTKV is different among PKD1 resolved and unresolved cases [P for interaction (age × PKD1 mutation) = .017] (Supplementary Table 1).

Kaplan–Meier renal survival analysis was performed to determine mean/median age to ESRD in PKD1 with truncated and non-truncated mutations (Fig. 5). Patients with PKD1 truncating mutations showed a shorter renal survival time {51.8 years [standard error (SE) 1.223]} compared with patients with non-truncating mutations [56.3 years (SE 0.918)], with statistical significance as indicate by the logrank test (P = .043) (Fig. 5 and Table 3). The overall mean age of ESRD, calculated by Kaplan– Meier renal survival analysis, for patients with PKD1 mutation was 53.9 years (SE 0.907), while the median was 55 years (SE 0.929) (Table 3).

Figure 5:

Kaplan–Meier renal survival analysis for ADPKD patients with PKD1 truncated and non-truncated mutations. Time to ESRD is plotted for each group.

Table 3:

Mean and median age of ESRD in patients with PKD1 mutations calculated by Kaplan–Meier analysis.

	Meana				Median
			95% confidence interval				95% confidence interval
Group	Estimate	Standard error	Lower bound	Upper bound	Estimate	Standard error	Lower bound	Upper bound
PKD1 truncated mutation	51.815	1.223	49.418	54.213	53	1.483	50.094	55.906
PKD1 non-truncated mutation	56.339	0.918	54.541	58.138	56	0.579	54.864	57.136
Overall	53.915	0.907	52.138	55.692	55	0.929	53.178	56.822

^aEstimation is limited to the longest survival time if it is censored.

DISCUSSION

This is the first report comprehensively describing the genetic landscape of ADPKD in Kuwait. We provide insights on the genotype–phenotype relationship in terms of disease severity evaluated by eGFR and htTKV. We recorded an overall molecular genetic diagnostic rate of 112/126 (88.9%) and 29/36 (80.6%) in families reporting 29 ADPKD pathogenic cases, of which 20/29 (69%) are novel and described for the first time. We highlight current limitations in the genetic diagnosis of ADPKD, as 19.4% of families were unsolved genetically.

Most pathogenic variants were detected in PKD1 (77.8%). Our results were supported by previous studies where PKD1 pathogenic variants were responsible for 67% and 77.7% of ADPKD cases in cohorts from Ireland and the USA, respectively [5, 27]. PKD1 exon 15 was the exon with highest number of pathogenic variants, but this result can be attributed to it genomic size (3620 bp) rather than being a mutation hot spot. In the PKD1 group, although the differences recorded in trends of eGFR decline and htTKV enlargement with age were not statistically significant between the PKD1 truncating and non-truncating groups, the more severe form of the disease in the PKD1 truncating group was demonstrated by the shorter time to ESRD, on average, when compared with patients with PKD1 non-truncating mutations (Fig. 4 and Table 3). At early stages of ADPKD, renal function remains stable and therefore eGFR may not be a reliable marker to monitor disease progression. On the other hand, a growing body of research has shown a progressive increase in TKV throughout the course of ADPKD, even at early disease stages where renal function is still intact, likely due to hyperfiltration of healthy nephrons [28–30]. Currently TKV is the preferred prognostic biomarker to monitor the progression of ADPKD and has been shown to predict the decline in eGFR [31]. It has been utilized as well as a surrogate marker for disease progression in ADPKD tolvaptan clinical trials [32].

The estimated difference between the median age to ESRD in PKD1 truncating versus non-truncating groups (53 versus 56 years) was more narrow than other reports (55 versus 67 years) [15]. This can be influenced by many factors, including the high prevalence of comorbidities in Kuwait and, in particular, diabetes and obesity [33, 34]. Several reports shed light on the relationship between genotype and disease phenotype, where PKD1 truncating mutations were associated with a more severe form of the disease, in terms of onset to ESRD, when compared with PKD1 non-truncating mutations and PKD2 mutations [5, 11, 15, 35]. However, a recent study showed that 18% of ADPKD patients with PKD1 truncating mutations experienced a milder form of the disease [36]. Taken together, these results suggest possible involvement of genetic/environmental factors that, at present, are not fully understood.

One of the rare cases diagnosed was a 26-year-old male who had a novel large 48.3-kb deletion that covered PKD1 exons 2–46 and extended to exons 18–41 from the TSC2, which lies immediately adjacent to PKD1 (tail to tail) on chromosome 16p (Table 2). This patient showed early renal sufficiency that overlapped with a spectrum of tuberous sclerosis manifestations. The TSC2-PKD1 contiguous deletion syndrome is a rare condition reported for the first time in 1994 [37]. Since then, ≈70 cases have been reported in the literature [38–40]. Earlier clinical intervention, facilitated by earlier molecular diagnosis, is likely to improve disease management and outcomes.

Genetic analysis and mutation screening in PKD1 can be challenging technically due to multiple reasons. First, the gene is highly polymorphic and, as we have shown, mutations of different types can occur along the entire length of the gene without mutation hot spots (Fig. 2). Second, ADPKD-linked mutations are likely to be unique to a single family, as previous reports have shown that recurrent pathogenic variants are reported in only 30% of the total detected ADPKD mutations [41], which we also found in our cohort, as novel detected pathogenic mutations accounted for 72.4% of the ADPKD mutations reported. Third, and most importantly, exons 1–32 of PKD1, are duplicated six times and located in pseudo-regions 13–16 Mb proximal to the genuine PKD1, with a shared sequence homology of up to 97.8% [42]. This region of PKD1 represents a diagnostic challenge, and particularly to WES, as it has been reported that this technique has shown low sensitivity in the detection of mutations in this region of PKD1, which could limit its utility in clinical setups [43].

PKD2 is the second common gene responsible for ADPKD, as its accounts for 15% of the cases [5, 6]. However, in our cohort we did not detect any pathogenic variants in PKD2 that are linked to ADPKD (Fig. 1). Mutations in PKD2 are usually associated with a milder form of ADPKD, with a mean onset of ESRD of 79.7 years [13], so patients with PKD2-linked disease might not seek or present late to medical attention, which may explain their underrepresentation in our cohort [13, 44]. Referral bias cannot be excluded.

While the majority of ADPKD cases in our cohort were attributed to PKD1, we characterized one ADPKD case linked to a mutation in IFT140, the gene recently described as the third common ADPKD gene (Fig. 1 and Table 2) [16]. We could not perform comprehensive clinical analysis of the phenotype associated with IFT140 because we have one singleton in our cohort, but the patients showed a milder form of the disease as indicated by the normal kidney function and kidney volume. This agrees with the findings of a recent report that indicated monoallelic mutations in this gene are associated with a milder form of ADPKD even when compared with PKD2-linked disease [16]. These findings highlight the degree of complexity in ADPKD diagnosis and pathophysiology that is also reflected in the wide spectrum of disease manifestation. This phenotypic variability might be useful in guiding molecular testing.

Intrafamilial phenotypic variability has been estimated to occur in at least 12% of families with ADPKD, which provides a strong indication for the involvement of modifier genes [45]. We reported previously a novel PKD1 variant (c.5305C>T, p.His1769Tyr) demonstrating a disease-modifying role in trans with a truncating PKD1 mutation (c.6727C>T, p.Gln2243*) in a family with a history of ADPKD where patients who co-inherited the variant and mutation showed larger renal volumes and earlier renal insufficiency when compared with those who only carried the pathogenic mutation (Table 2) [46]. We further identified a total of nine variants, five of which were co-inherited with ADPKD pathogenic mutations in singleton cases (Table 2). The co-inheritance of certain variants in genes with ADPKD relevance can contribute to worse disease symptoms, as we reported in a case where a 42-year-old patient co-inherited a truncating large PKD1 deletion (c.3656_6911del3256) along with a VUS PKD2 splice site variant (c.2713A>C) (Table 2). The patient showed a more severe form of the disease demonstrated by an early onset of ESRD (39 years old). While understanding the exact prognostic effect of such genetic modifiers can help in improving disease management, it can be challenging, at times, and might require larger family-based cohort studies.

In our cohort, 19.4% of family cases were classified as unsolved, as no mutations linked to ADPKD were identified (Fig. 1), which is close to the 16% reported in Ireland [27]. Other studies reported 5–10% of unsolved cases in their PKD cohorts [15, 17, 47]. Despite the variable proportions between cohorts, which can be attributed to the design of the genetic testing pipeline, ADPKD genetically unsolved cases remain a diagnostic challenge that can limit prognostic application of genetic testing and affect any proposed family planning options. Some of these cases could be a result of genetic PKD1/PKD2 mosaicism, and conventional genetic analysis seems incapable of identifying such mutations [6]. The involvement of new genes other than the ones already linked to ADPKD is also a possibility. This genetic heterogeneity is also a contributor to the variable spectrum of ADPKD disease manifestations and severity. Our results indicated no apparent association between htTKV and age in our unresolved cases (Fig. 4D). This could be explained by the possible high degree of genetic heterogeneity of this particular group and subsequent variable disease manifestations.

ADPKD genetic testing can improve the care of patients by guiding therapy, especially when integrated with other clinical parameters such as gender, hypertension and urological events to predict disease progression and aid personalized therapeutic decisions [48]. Genetic testing can also be useful to include/exclude patients for certain treatments, for example, excluding PKD1/2-negative ADPKD patients from tolvaptan treatment [49, 50].

Apart from conventional general CKD care measures, the current standard of care provided for ADPKD patients in Kuwait focuses mainly on managing the disease complications. While this may limit morbidity and mortality due to ADPKD-related complications, it does not delay disease progression. The current available treatment option for rapidly progressive ADPKD patients reaching ESRD in Kuwait is renal replacement therapy, which is placing a significant burden on ESRD care facilities. In a recent study on the dialysis population in Kuwait, PKD was shown to account for ≈2% of the total dialysis population [51]. However, unpublished data on renal transplants in Kuwait show that ADPKD cases accounted for 9.73% of total renal transplants performed between 2016 and 2019 (Supplementary Table 3). In 2018 the US Food and Drug Administration authorized use of the vasopressin V2 receptor antagonist tolvaptan as the first drug treatment for ADPKD [32]. Clinical trials showed that the use of tolvaptan reduced the growth of kidneys and slowed renal function decline in adult ADPKD patients at risk of rapid disease progression [32, 52, 53]. In Kuwait, tolvaptan use for ADPKD patients is due to start in 2022.

The presented study has some limitations. The relatively small cohort size and subsequent smaller subgroup sizes could restrict direct informative statistical comparisons. Some of the clinical data were self-reported by the recruited patients, thus the possibility of self-reporting bias cannot be excluded. There was also an underrepresentation of PKD2 cases, possibly caused by referral bias. Using ultrasonography to measure kidney volume is operator dependent and relies on geometric assumptions that could affect its accuracy when compared with CT or MRI.

CONCLUSION

The observed genetic heterogeneity of ADPKD along with the wide phenotypic spectrum reveal the level of complexity in disease pathophysiology. ADPKD genetic testing could improve the care of patients through improved disease prognostication, treatment guidance, support of clinical trials and genetic counselling. However, it is also important to overcome the hurdle of genetically unresolved ADPKD cases thorough improved genetic testing pipelines to ensure more effective implementation of ADPKD genetic testing in clinical setups.

FUNDING

This work was funded by the KFAS research fund (PR17-13MM-07, to H.A.) and the National Institute of Diabetes and Digestive and Kidney Diseases (grant DK058816, to P.H.).

AUTHORS’ CONTRIBUTIONS

H.A. was responsible for conceptualization, data curation, funding acquisition, the investigation, project administration and writing the manuscript. M.N., A.AlSahow, Y.B., A.Al-Hunayan, N.H., M.Al-Ali and M.AlMousawi were responsible for patient recruitment and clinical analysis. S.R.S. was responsible for genotyping, data analysis and review and editing of the manuscript. M.A. and J.A. were responsible for resources and the investigation. A.M. was responsible for data analysis and visualization. A.M.A. and M.Z. were responsible for radiological analysis. S.D. was responsible for patient recruitment. S.E.J. and A.C. were responsible for genetic data analysis. T.A.T. was responsible for data interpretation. F.A. was responsible for supervision and data interpretation. P.C.H. was responsible for conceptualization, data analysis, the investigation, resources and review and editing of the manuscript.

DATA AVAILABILITY STATEMENT

The data underlying this article will be shared upon reasonable request to the corresponding author.

CONFLICT OF INTEREST STATEMENT

The authors declare no competing interests.