Fourteen Monogenic Genes Account for 15% of Nephrolithiasis/Nephrocalcinosis

• kidney stones
• hypercalciuria
• human genetics
• molecular genetics
• renal tubular acidosis
• Bartter syndrome

Nephrolithiasis (NL) is a highly prevalent condition affecting up to 10% of individuals in the Western world.¹ It is associated with significant morbidity because of colicky pain, the necessity of surgical procedures, and progression to CKD.² NL and related conditions, such as nephrocalcinosis (NC), share a well recognized heritability, emphasized by the fact that up to two thirds of hypercalciuric stone formers have relatives with NL.^3–5 There are at least 30 genes that have been shown to cause single-gene forms of NL/NC by autosomal-dominant, autosomal-recessive, or X-linked transmission.^5,6 The ability to detect the causative mutation(s) in a monogenic disease gene is of great diagnostic and potentially therapeutic importance, because there is an almost deterministic cause–effect relationship in monogenic disease.

However, the contribution of monogenic disorders to the overall prevalence of NL has never been studied comprehensively. Hence, data are lacking on the frequency of monogenic forms of NL in the overall population of stone formers. Furthermore, absence of a positive family history, which will be the rule in recessive genes and may be frequent in dominant genes with incomplete penetrance, often leads to the false assumption that a monogenic cause is unlikely. We, therefore, hypothesized that monogenic causes of NL/NC account for a high percentage of stone formers. For most individuals with NL/NC, mutation analysis for a causative genetic defect has not been accessible so far. With the advent of massively parallel sequencing techniques and high-throughput library preparation, mutation analysis of multiple genes in large cohorts has not only become technically feasible but also, cost-effective.⁷

To determine the percentage of monogenic causes in kidney stone disease, we analyzed, in a cohort of 272 genetically unresolved individuals (268 families) (Supplemental Table 1), all coding exons and splice sites of 30 known NL/NC-causing genes with a defined Online Mendelian Inheritance in Man (OMIM) phenotype (Supplemental Table 2 and Concise Methods). Participants were recruited consecutively and without prior preselection at three typical kidney stone clinics on the basis of diagnosed NL or NC as outlined in Concise Methods. We applied a high-throughput mutation analysis technique that we recently developed.^7,8 As a result, we show that causative mutations in known NL/NC-causing genes are present in around 15% of these 268 families, and we suggest possible therapeutic and prophylactic implications for some of these genetic findings.

By targeted resequencing of 381 coding exons of 30 genes known to cause autosomal-dominant, autosomal-recessive, or X-linked NL/NC if mutated, we identified 12,589 single-nucleotide variants and 1246 deletion-insertion variants in 336 individuals (272 genetically unresolved individuals and 64 controls). To distinguish benign variants from disease-causing mutations, we evaluated each variant individually on the basis of strict criteria as described in Concise Methods. Overall, 162 (1.2%) variants met our criteria for being likely disease-causing variants, 139 of which were confirmed by Sanger sequencing (86%), whereas the others represented low-representation artifacts of multiplex PCR. After exclusion of single heterozygous variants in recessive genes and nonsegregating variants and subtraction of variants from duplicates and positive controls, a total of 50 most likely disease-causing variants remained (Supplemental Table 3).

As a result, we made a causal molecular genetic diagnosis that is likely to explain the disease phenotype in 40 unrelated individuals in our total cohort of 268 families with NL/NC (14.9%) (Tables 1 and 2). Pathogenic mutations were detected in the following six dominant disease genes: SLC7A9 (19 families), ADCY10 (3 families), SLC2A9 (2 families), SLC9A3R1 (2 families), SLC22A12 (2 families), and SLC4A1 (1 family) (Table 1, Supplemental Table 4). Furthermore, we identified pathogenic mutations in the following eight recessive disease genes: SLC3A1 (two families), CLCN5 (two families), CLDN16 (two families), ATP6V1B1 (one family), CYP24A1 (one family), AGXT (one family), SLC34A1 (one family), and SLC34A3 (one family) (Table 2, Supplemental Table 4).

No pathogenic mutations were identified in the genes APRT, ATP6V0A4, CA2, CASR, CLCNKB, CLDN19, FAM20A, GRHPR, HNF4A, HOGA1, HPRT1, KCNJ1, OCRL, SLC12A1, VDR, and XDH. In total, 20 of 50 mutations (40%) detected were novel pathogenic variants, which have not been previously reported (Human Gene Mutation Database; http://www.hgmd.cf.ac.uk/ac/index.php).^9–22

In the pediatric subgroup of individuals with age of onset before 18 years (defined by age at first stone or age at diagnosis of NC), we identified a causative mutation in 20.8% (22 of 106) of individuals (Figure 1A). In contrast, in the adult cohort (defined by age of onset ≥18 years), deleterious variants were detected in 11.4% (19 of 166) of individuals (Figure 1A). The distribution of median age of onset across 14 genes, in which causative mutations were found, correlated with the mode of inheritance as follows: although four of six dominant genes were associated with adult manifestation (≥18 years), mutations in recessive genes were mainly identified in probands with an age of onset before 18 years (Figure 1B).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.

Distribution of established molecular genetic causes of NL/NC. (A) Percentage of subjects with identified molecular diagnoses across grouped ages of onset. This percentage is significantly higher in the pediatric cohort (age of onset<18 years) compared with the adult cohort (age of onset≥18 years): 20.8% versus 11.4%. *P≤0.05. In A and B, the green rectangles indicate six individuals with heterozygous SLC7A9 mutations enriched within the age group ≥18–30 years (6 of 19=31.6%). (B) Distribution of age of onset across mutated causative genes. Genes with a dominant mode are annotated in black, whereas genes with a recessive mode are annotated in red. ^†For SLC2A9, SLC22A12, and SLC7A9, detected mutations were primarily dominant, although both modes of inheritance have been reported. Detected mutations in four of six genes with a median onset ≥18 years are in dominant genes (upper right quadrant as indicated by dashed lines), whereas six of eight genes with a median onset of disease <18 years are in recessive genes (lower left quadrant as indicated by dashed lines).

For the gene SLC7A9, which was found most commonly mutated in this cohort, there was no correlation between age of onset and allelic strength (truncating versus missense variant) (Supplemental Figure 1A). Almost one third of individuals with SLC7A9 mutations manifested within the age range of ≥18–30 years, with a median at 26 years (Figure 1). However, there was a genotype–phenotype correlation in that individuals with two pathogenic variants of SLC7A9 (8 of 19 unrelated individuals) showed a significantly earlier manifestation compared with individuals with only one pathogenic variant (mean: 18 versus 30 years) (Supplemental Figure 1B). For approximately 60% of individuals, the identified genetic diagnosis confirmed the previously obtained clinical diagnosis. In approximately 40% of cases, however, the genetic diagnosis contributed additional etiologic and diagnostic information to what was clinically suspected, suggesting practical implications (Tables 1 and 2).

We here examined an international cohort of 272 typical kidney stone formers for the presence of mutations in 30 genes that cause NL/NC if mutated. We identified 50 (20 novel) pathogenic alleles in 14 different genes in 40 of 268 families (14.9%).

This work, to the best of our knowledge, is the most extensive genetic screening of known NL/NC-causing genes in a combined cohort of pediatric and adult individuals with kidney stones and/or NC. The overall percentage of families with pathogenic mutations exceeds the general assumption of a relatively small contribution of monogenic causes to the general population of stone formers. Remarkably, in almost 21% of the pediatric cohort and 11% of the adult cohort, we identified causative mutations in 1 of 14 genes. The fact that we did not find mutations in the remaining 16 genes suggests that mutations in those genes are less prevalent. Although it is generally assumed that around 85% of causative mutations in monogenic disorders reside within coding regions and adjacent splice sites,²³ copy number variations and deleterious deep intronic variants are undetectable with the screening approach that we applied to this study. These limitations as well as population genetic factors may have led to false negatives and a selection bias in regards to the distribution of molecular diagnoses in this study cohort.

SLC7A9 was, by far, the most prevalent disease-causing gene in our cohort, with a median age of first stone at 26 years (Figure 1B). This finding is in line with retrospective data derived from stone composition analysis showing the predominance of cystinuria as the major monogenic cause of stone disease in the adult population.²⁴ Presence of two pathogenic variants in SLC7A9 led to earlier manifestation in our cohort (Supplemental Figure 1B). Whether this finding is because of recessive inheritance or the presence of two mutations on the same allele could not be fully investigated for reasons of incomplete allelic segregation data. Interestingly, in six individuals with mutated SLC7A9, clinical data did not raise the suspicion of cystinuria, and three of them exhibited calcium-containing kidney stones (JAS-C8, JAS-E5, and JAS-F87) (Tables 1 and 2). There are recent reports showing similar findings of calcium-based stones in individuals with heterozygous SLC7A9 mutations.^25,26 Additionally, Martins et al.²⁷ showed that cystine promotes calcium oxalate crystal formation in vitro, implying that cystinuria may be a risk factor for calcium oxalate calculi. In conclusion, these data emphasize the need to screen urine for the presence of excess dibasic amino acids, including cystine, in all stone formers, even if the molecular genetic diagnosis has been established.²⁸ Although cystinuria is the most common clinically known genetic diagnosis, in some of these individuals, only broad genetic screening revealed the etiology, affecting their future treatment and the preventative measures that could then be instituted.

Admittedly, for ADCY10, in which we found three putatively deleterious alleles, the evidence for single-gene causation is controversial. The OMIM phenotype implies susceptibility to absorptive hypercalciuria. This notion is on the basis of a single case control study by Reed et al.²⁹ and has never been confirmed in larger-sized cohorts.

Nevertheless, our study supports the conception that NL/NC is genetically broadly heterogeneous. We provide evidence that the role of monogenic causes accounts for a significantly higher percentage of kidney stone formers than generally suspected, especially in the adult population. Most importantly, identification of the monogenic causes of NL/NC may not only have important prognostic implications but also lead to therapeutic consequences (Tables 1 and 2): in >40% of the cases in our study, the molecular genetic diagnoses contributed a new aspect to the previously established clinical diagnosis, suggesting practical implications, such as avoiding vitamin D (CYP24A1), initiating audiometry (ATP6V1B1), or excluding the risk of recurrence in renal transplants (CLCN5 or CLDN16) (Tables 1 and 2).

An excellent example of individualized therapy on the basis of molecular genetic diagnostics was previously shown for individuals with primary hyperoxaluria type 1 caused by AGXT mutations, where pyridoxine sensitivity is associated with the presence of a distinct allele (Gly170Arg).³⁰ Furthermore, invasive and potentially harmful procedures, such as a diagnostic liver biopsy in individuals with suspected primary hyperoxaluria type 1, can be circumvented after diagnostic gene panels become part of the clinical repertoire.³¹ The use of such a broad genetic screening device may also help raise awareness of extremely rare disorders and be beneficial in cases of atypical clinical presentation or hindered standard diagnostics because of advanced CKD.

• Copyright © 2015 by the American Society of Nephrology