Phenotypic Discordance among Siblings with Autosomal Recessive Polycystic Kidney Disease: Case Report and Review of the Literature

Autosomal recessive polycystic kidney disease (ARPKD) is defined by a characteristic pattern of renal collecting duct dilation and hepatic ductal plate malformation with periportal fibrosis. The proportion of affected renal tubules and the degree of hepatic involvement exhibits broad variation between patients. The disease typically has neonatal or childhood onset but can present in youth or early adulthood in a minority of patients [1]. Early studies showed phenotypic concordance between affected siblings, which led to the nosological splitting of ARPKD into discrete genetic entities based on disease severity and age of onset [2]. However, in 1988, Kaplan et al. [3] analyzed kindreds where one sibling had severe perinatal-onset disease while another had mild juvenile-onset disease. Their findings supported a return to the single gene concept, which was later confirmed by the positional cloning of PKHD1 [4].

While biallelic truncating variants all but invariably lead to severe perinatal-onset disease, missense PKHD1 variants are associated with the full spectrum of disease severity and exhibit variable expressivity [5]. It has been proposed that disease-modifying alleles could be found at other loci associated with a polycystic phenotype such as DZIP1L, HNF1B, PKD1, PKD2, and GANAB [6, 7]. But a systematic characterization of candidate modifier genes in a highly discordant sibship has not been reported.

In this report, we describe a discordant sibship composed of a proband who presented with progressive chronic kidney disease (CKD) and Caroli syndrome in childhood requiring combined liver and kidney transplantation and his sister who is minimally affected at age 33 with signs of medullary sponge kidney. The patients inherited the PKHD1 c.4870C>T (p.Arg1624Trp) variant from their father and the c.8206T>G (p.Trp2736Gly) variant from their mother (Fig. 1a). Next-generation sequencing of genes associated with cystic kidney disease did not identify additional likely pathogenic variants at loci that have been hypothesized to act as modifiers. We review previously reported cases of phenotypic discordance among siblings and suggest that genotypes composed of at least one hypomorphic missense variant are more likely to lead to phenotypic discordance.

At 1 year of age, patient A, a male, was brought for medical assessment due to skin findings suggestive of petechiae. He was found to have bilateral polycystic kidneys as well as cystic dilation of intrahepatic bile ducts diagnosed as Caroli syndrome. The size of the kidneys at that time is not known to us. His kidney function progressively declined over the years, and he eventually reached kidney failure at age 19 requiring hemodialysis. A kidney transplant at age 20 failed in the context of repeated infections likely secondary to cholangitis due to Caroli syndrome, and he returned to hemodialysis at age 24. An MRI at age 26 showed small kidneys with numerous cysts and evidence of significant periportal fibrosis and portal hypertension with splenomegaly, moderate ascites, and gastrosplenic varices (Fig. 1b, c). The patient also has a history of systemic hypertension, nonischemic cardiomyopathy with an ejection fraction of 40–45%, and secondary hyperparathyroidism with signs of renal osteodystrophy. At age 31, he received a combined liver and kidney transplant.

Patient B, the sister of patient A, was first seen by a nephrologist at age 26 in the context of a symptomatic kidney stone. The kidney ultrasound and CT scan showed an atrophic left kidney (9.1 cm) and compensatory hypertrophy of the right kidney (14.2 cm) with bilateral nephrocalcinosis and several simple kidney cysts in the right kidney. A subsequent MRI was suggestive of medullary sponge kidney and showed some small liver cysts (Fig. 1d). At the most recent follow-up (33 years old), her kidney function remained normal, but she developed moderately increased albuminuria (36 mg/mmol) and her latest kidney ultrasound was unchanged. Her liver transaminases, liver function tests, and serum bile acids all remained normal throughout her follow-up. Apart from the single episode of kidney stone, her only symptoms are slight polyuria and polydipsia.

Patient A underwent clinical genetic testing for monogenic causes of cystic kidney disease with a comprehensive kidney disease panel and was found to have biallelic likely pathogenic PKHD1 variants, c.4870C>T (p.Arg1624Trp) and c.8206T>G (p.Trp2736Gly) (Table 1, Fig. 2). The gene panel did not identify additional likely pathogenic variants or variants of uncertain significance (VUS) in cystic kidney disease genes that have been hypothesized to act as modifiers (online suppl. Table 1; for all online suppl. material, see https://doi.org/10.1159/000540741). Four VUS were found in three genes associated with kidney diseases that do not correspond to the patient’s phenotype, namely, HPSE2 (urofacial syndrome 1), SCCN1A (pseudohypoaldosteronism type 1), and SLC4A4 (proximal renal tubular acidosis with ocular anomalies, 2 VUS). Genotyping of the parents confirmed that the PKHD1 variants are in trans. Carrier testing of seemingly unaffected family members identified both PKHD1 pathogenic variants in one of the proband’s sisters. Further testing of the sister on an exome backbone did not identify pathogenic variants in other kidney disease genes that may offer a competing explanation for her divergent phenotype. The sister was found to share the VUS in HSPE2 and SLC4A4 with her brother and to carry heterozygous VUS in DLC1 (proteinuric kidney disease), CSPP1 (Joubert syndrome), and INF2 (focal segmental glomerulosclerosis). The patients are siblings from a non-consanguineous Syrian couple (Fig. 1a). The mother is a healthy kidney donor, and the father has a history of polyuria and two episodes of kidney stones. The father’s kidney function, urinalysis, and kidney stone workup including random urine-specific gravity, oxalate-creatinine ratio, and calcium-creatinine ratio were within normal limits. A non-contrast CT scan of the father showed two small simple kidney cysts and no signs of nephrocalcinosis. An older sister, aged 36, is a carrier of the p.Arg1624Trp variant inherited from the father and has a history of polyuria.

We tabulated the genotypes of discordant sibships that have been previously reported in the literature (Table 2) [8‒14]. We used the pathogenicity prediction tool AlphaMissense to compare the missense genotypes carried by sibships with discordant phenotypes to those reported in a large ARPKD reference cohort from ARegPKD and RWTH Aachen University [15, 16]. Interestingly, we found that all 18 biallelic missense genotypes from discordant sibships are composed of at least one allele that is considered ambiguous or benign by AlphaMissense (Fig. 3a). In contrast, the reference ARPKD cohort includes 13 out of 88 missense genotypes where both variants are considered likely pathogenic by AlphaMissense (Fig. 3b). Due to the small sample size of discordant sibships, the differences in predicted pathogenicity class between the discordant cohort and the reference cohort fall short of statistical significance at the 0.05 level (predicted pathogenic/pathogenic class: one-tailed Fisher’s exact test, p = 0.075). In addition, we noted that the proband genotype falls in the predicted pathogenic/benign category along with four previously reported genotypes that have been associated with discordant sibships, R1624W/I2957T, R1624W/N3063K, P1580S/I1638T, and W656C/G1123S (Table 2). When plotting the location of variants associated with discordant sibships along the PKHD1 protein, we noted the lack of variants in the PA14 and the second G8 domain, as opposed to the cluster of variants in these domains observed in the reference ARPKD cohort (Fig. 3c).

We describe non-consanguineous siblings with biallelic pathogenic missense variants in the PKHD1 gene who display highly discordant phenotypes. Notably, while patient A presented with cystic kidney disease and Caroli syndrome in childhood and developed kidney failure in the second decade of life, patient B has normal kidney function and reports minimal symptoms, such as one kidney stone episode, in the fourth decade of life. In addition to atrophy of the left kidney and simple kidney cysts in the right kidney, imaging of patient B showed signs of medullary sponge kidney, a finding which has been previously described in some patients with adult-onset ARPKD [17, 18]. In fact, the characteristic dilation of the terminal collecting ducts seen in medullary sponge kidney bears resemblance to that found in ARPKD [19]. This tubular ectasia leads to nephrocalcinosis and increased risk of kidney stones. A study of parents of patients with ARPKD found that some obligate heterozygous carriers had increased medullary echogenicity on ultrasound suggestive of medullary sponge kidney [20]. Though this may offer an explanation for the history of kidney stones in the patient’s father, a non-contrast CT scan did not show signs of medullary nephrocalcinosis. Among the previously reported discordant sibships with known pathogenic genotype, this report is one of the first to describe a minimally affected sibling in the fourth decade of life. This appears highly unusual for an autosomal recessive disease that is generally considered to have childhood onset. It has been suggested that phenotypic discordance could be caused by additional variants in other polycystic kidney disease genes; however, we could not confirm this for our patient.

The study of expressivity is limited by the extensive allelic heterogeneity within the PKHD1 gene, which encodes a 4,074-amino-acid protein (online suppl. Fig. 1). Variable expressivity has been described among affected siblings as well as nonrelatives homozygous for founder variants (e.g., c.1880T>A [p.Met627Lys] in the Afrikaner population [9]). A cohort study of 48 families with at least one neonatal survivor found that for 20 sibships one sibling developed severe disease leading to death in the perinatal or neonatal period while another presented in childhood or youth; all discordant sibships carried at least one missense variant [8]. This pattern was also exhibited by siblings homozygous for the Afrikaner founder variant [9]. Recently, the ARegPKD registry study reported that, among a cohort of neonatal survivors, three sibling pairs out of twenty developed discordant phenotypes with one sibling showing stage G1 or G2 CKD and the other kidney failure at a similar age [10]. Two of these discordant pairs carried biallelic missense variants, while the genetic testing of the third was not reported. A similar finding was previously reported in a cohort study of 73 patients: one sibling with biallelic missense variants required kidney transplantation at age 18 and had systemic hypertension at age 1, while three other affected siblings had stage G1 or G2 CKD in the third decade without hypertension [11].

An outstanding question is to determine which PKHD1 genotypes are strongly susceptible to the effects of the genetic background and environmental factors. We noted that our patients’ genotype is composed of a hypomorphic missense variant, p.Arg1624Trp, and a strongly disruptive missense variant, p.Trp2736Gly. The p.Arg1624Trp variant has a reported allele frequency of 0.015% in the gnomAD exome global population, which is greater than most pathogenic PKHD1 alleles (online suppl. Fig. 1), and may have a higher allele frequency in Middle Eastern populations (0.19% in the Turkish Variome) [21]. The variant has been reported in patients with a milder, later-onset phenotype though some compound heterozygous individuals with perinatal onset have also been reported [11, 22‒25]. In a cohort of consanguineous CAKUT patients, an individual with renal hypodysplasia and kidney failure was found to be homozygous for p.Arg1624Trp; subsequent liver phenotyping identified Caroli disease and gallbladder stones [26]. A 20-year-old patient who presented with bilateral medullary sponge kidneys and later developed CKD rapidly progressing to kidney failure was found to be compound heterozygous for p.Arg1624Trp and p.Leu2505Ter [18]. A discordant sibship was reported to be compound heterozygous for p.Arg1624Trp and p.Ile2957Thr; the older brother was found to have liver cysts and CKD at 2 months of age progressing to stage G3a at age 15, while the younger sister remained asymptomatic at age 12 with preserved kidney function and no signs of cysts on ultrasound [12]. The impact of p.Trp2736Gly on protein function is likely to be more severe than p.Arg1624Trp given its low allele frequency, consistently pathogenic scores from in silico prediction tools and high degree of evolutionary conservation (Table 1; Fig. 2). A compound heterozygous individual for p.Trp2736Gly and p.Thr36Met died in the neonatal period and had enlarged kidneys and congenital hepatic fibrosis on autopsy [27]. Two other compound heterozygous individuals were reported in the literature: one with p.Arg3240Leu who presented prenatally [22] and another with a truncating frameshift variant who died in the neonatal period [28].

When reviewing the literature for commonalties between our sibship’s genotype and those reported in other sibships with discordant phenotypes, we noted that all reported genotypes are composed of at least one variant predicted to be benign or ambiguous by AlphaMissense (Fig. 3a). Since AlphaMissense scores are not influenced by human classification found in curated databases like ClinVar, we used the tool as an independent indication of the severity of the impact of pathogenic missense variants on protein structure, allowing us to distinguish strongly disruptive missense variants from hypomorphic missense variants [16]. Our findings suggest that, for each genotype previously reported to show phenotypic discordance among siblings, at least one variant exerts hypomorphic effects. In other words, the presence of a hypomorphic missense variant is required for a genotype to show significant variable expressivity. Interestingly, up to 85% of the missense genotypes reported in a reference ARPKD cohort consist of at least one hypomorphic variant according to our interpretation of AlphaMissense predictions; we could therefore expect a proportion of these genotypes to show variable expressivity (Fig. 3b). We note that the most common PKHD1 pathogenic variant, c.107C>T (p.Thr36Met), is classified as ambiguous by AlphaMissense. It has been proposed based on in silico prediction with NetStart that the variant exerts its pathogenic effect by creating an alternative translational start site resulting in a truncated PKHD1 protein [23]. Since AlphaMissense scores amino acid substitutions according to their probable effect on the structure of a defined protein isoform, the tool cannot account for the pathogenic effect of changes in translational start site usage created by missense variants. However, this limitation is unlikely to affect the prediction accuracy of most pathogenic missense variants. Finally, no phenotypic discordance among siblings has yet been reported for variants in the PA14 domain and the second G8 domain; however, this observation may be biased due to the small number of reported discordant sibships (Fig. 3c). The PA14 domain is found in diverse bacterial and eukaryotic proteins including signaling proteins, where it serves to bind carbohydrate-containing ligands [29]. The second G8 domain contains five repeated β-strand pairs and may also be involved in extracellular ligand binding [30]; data from the reference ARPKD cohort indicate that it is associated with poorer hepatic outcomes [15].

This report with its associated review of the literature suggests that, when counseling siblings with pathogenic genotypes composed of at least one hypomorphic PKHD1 missense variant, the clinician must be aware of the possibility of significant phenotypic discordance. Further reports of variable expressivity are needed to determine the characteristics that render certain PKHD1 pathogenic genotypes susceptible to the effects of the environment and possible genetic modifiers.

Written informed consent was obtained from the patients for publication of this case report and any accompanying images. This study protocol was reviewed and approved by the McGill University Health Centre Research Ethics Board (Approval No. 2023-9387).

The authors have no conflict of interest to declare.

This study was not supported by any sponsor or funder. Rémi Goupil is a clinical research scholar from the Fonds de recherche du Québec-Santé. Thomas M. Kitzler is funded through support by the Fonds de recherche du Québec en santé (FRQS) Salary Grant and the SickKids New Investigator Research Grant.

Marc Henein and Thomas M. Kitzler conceptualized and designed the study. Marc Henein, Felicia Russo, Rémi Goupil, and Thomas M. Kitzler obtained the clinical information and contributed to data interpretation. Zachary T. Sentell contributed to the literature review and data interpretation. Marc Henein performed the sequencing and genotypic analysis and wrote the manuscript with support from Thomas M. Kitzler. Felicia Russo, Zachary T. Sentell, and Rémi Goupil reviewed the manuscript.

Publicly available datasets were used in this study. These can be found in [15, 16].