Since the last quarter of the 20th century, genetic testing has been increasingly used in clinical practice. In most situations, a genetic test is used to identify the precise genetic change responsible for a Mendelian disease in an individual or a family. Identification of such a “molecular diagnosis” can have numerous impacts on care or decision making for a patient or their wider family:

• 1.

  It can confirm the precise diagnosis, sometimes obviating the need for invasive investigations or a prolonged diagnostic odyssey and sometimes confirming the mode of transmission (including which relatives are potentially at risk);

• 2.

  It can provide clarity for a patient or family as to why they have developed a disease, and this can be psychologically helpful, although it can also lead to feelings of guilt among parents who may feel “responsible” for passing on a genetic change;

• 3.

  It can provide prognostic information, which in turn can help with life choices such as career or family planning (including a decision to adopt or not have children), can help decide whether treatment is indicated, and sometimes it can reveal the risk of disease recurrence following transplantation;

• 4.

  It can inform selection of specific therapy or obviate the need for therapies known to be ineffective (such as corticosteroid treatment for monogenic podocytopathy presenting with focal segmental glomerulosclerosis);

• 5.

  It can allow predictive testing to be offered to at-risk individuals. This, in turn, can:

  • a.

    inform their key life decisions about career, financial, or family planning (whether or when to have children, or whether to use reproductive interventions);

  • b.

    allow them to access health or life insurance that would otherwise be more expensive or unavailable (with a negative test in the setting of a positive family history, noting that in some jurisdictions (e.g., Australia and UK) a positive predictive test result is privileged information that need not be shared with insurers);

  • c.

    allow access to (or discharge from) a surveillance program aimed at mitigating their risk from the disease;

  • d.

    inform transplantation decisions, allowing or precluding safe donation of a kidney by an ostensibly unaffected relative;

  • e.

    However, even where clinically or otherwise helpful or actionable, predicting a diagnosis in an otherwise healthy individual can have negative psychological consequences, sometimes characterized by patients with pervasive or unpleasant imagery of a “ticking time bomb,” It is important to address this possibility before genetic testing where possible (e.g., see https://www.nhs.uk/conditions/predictive-genetic-tests-cancer/ ). This dilemma will be further highlighted and will need to be resolved in a scholarly and thoughtful manner as more groups are offered genome sequencing at a population level and genetic variants are revealed.

• 6.

  A predictive test can be used for reproductive interventions, including prenatal or preimplantation genetic diagnosis, allowing a couple naturally at high risk of having an affected child the opportunity to have an unaffected child.

The utility and impact (medically, psychologically, and socially) of a molecular diagnosis in an individual or family can therefore be enormous, and there has been much effort in delivering genetic testing to patients for this reason: There are numerous examples across the medical literature and popular press justifiably extolling the benefits of genetic testing. However, because of the extremely consequential decisions that patients, relatives, and clinicians can sometimes make on the basis of a genetic test result, it is absolutely imperative that the information yielded by such a test is accurate and correctly interpreted by the clinician. Erroneously actioning a genetic variant that confers only a low (or even zero) risk of a disease as if it is “the cause of disease” in an individual can have negative or even catastrophic consequences, both medical (such as unnecessary mastectomy in a young woman harboring only harmless variants in a BRCA gene, or mistakenly precluding donation of a kidney by a healthy parent to their child on dialysis owing to detection of a low- or no-risk genetic variant) or for the patient’s wider life decisions (such as a decision not to have children, or not to embark on a particular career because of the incorrect belief that they [or their children] will become unwell later in life).

Historically (i.e., before the advent of massively parallel sequencing technologies), ascertainment of rare genetic variants was expensive, time consuming, and not widely available to nephrologists or other non-geneticists, so it was limited to those cases where there was a high prior probability of underlying Mendelian disease caused by a change in the gene (or few genes) analyzed. This meant that observation of a novel variant in such a targeted test performed in such a high-risk individual could be clinically consequential. Since the precipitous drop in sequencing cost and now widespread availability of large gene panels or even whole-exome sequencing in individual patients (termed “genomic testing”), the prior likelihood of a change seen in any one gene actually being the cause of a patient’s disease has dropped. This is compounded by the fact that it is increasingly recognized that rare genetic variants can be associated with nonsyndromic (e.g., renal-limited) forms of diseases previously regarded as distinct syndromes. One example is the LMX1B variants associated with proteinuric kidney disease and focal segmental glomerulosclerosis (FSGS) in the absence of abnormal nail or patella development that defined the Nail-patella syndrome, in which variants in this gene were originally identified. Together, this means that the previous situation where genetic tests would only be used after a clinical diagnosis of a specific syndrome or disease had been made is now being replaced by the paradigm in which larger numbers of patients lacking a specific clinical diagnosis are now undergoing large-scale genomic analysis. This has yielded a key benefit of securing an actionable molecular diagnosis in many more patients (as evidenced by the 15%–20% frequency of genetic diagnoses reported in patients with previously unexplained nephropathies undergoing exome sequencing ). But, as Bayes theorem indicates, the lower prior probability of a particular gene being responsible in patients undergoing testing has a dramatic effect on the posterior probability (and hence positive predictive value) of an identified genetic variant being the cause of a patient’s disease. This mandates, more than ever before, that genetic variants identified in a patient are rigorously appraised to ensure that the actionability of each variant is commensurate with the evidence causally linking it to disease.

Understanding how to interpret (and therefore when to perform) genetic and genomic testing in the modern era therefore requires knowledge of not just which phenotypes or presentations can be associated with which types of genetic change (i.e., the qualitative question of whether a genetic change identified in a patient perturbs the function of a protein in a way consistent with their disease), but it also requires a quantitative understanding of how strongly a given type of genetic change is related to a disease (i.e., what is the absolute risk of the disease attributable to the genetic change identified in a patient). In communicating genetic test results to a patient, the clinician therefore needs to understand what information, for the variant in question, constitutes strong enough data on which to base potential highly consequential clinical and life decisions. This chapter therefore addresses the types and distributions of genetic variants found in humans; the clinical indicators of, and biological mechanisms causing Mendelian and non-Mendelian disease, as well as the sources of evidence linking genetic variation with disease. Finally, specific clinical presentations encountered by nephrologists are discussed to illustrate how this knowledge can inform clinical practice ( Box 44.1 ).

The frequency of the less common (or minor) allele of each variant can range between 0 and 0.5, with the number of variants in a typical person shown in ( Table 44.1 ). However, the number of variants in each frequency category shows marked variation across different human populations, with the number of rare variants per individual being notably higher among those with recent African ancestry.

Looked at across a population ( Fig. 44.1 ), the vast majority of variants that can be detected by genome sequencing are ultrarare or private (i.e., occur in a single individual).

The minor allele frequency (MAF) distribution in a European population.

Reproduced from Ameur A, Dahlberg J, Olason P, et al. SweGen: a whole-genome data resource of genetic variability in a cross-section of the Swedish population. Eur J Hum Genet . 2017;25(11):1253–1260.

In autosomal recessive diseases, the presence of two defective and no normal alleles of a relevant gene is usually necessary and sufficient to cause disease, and in general, it is deficiency (complete or near complete) of the normally functioning protein that leads to disease. This may be due to the inheritance of two identical defective alleles, called homozygosity, or two heterozygous alleles, termed compound heterozygosity when the alleles are in trans (i.e., there is a fault in both copies of the gene). Because a single normal copy of the gene is usually enough to prevent disease manifestations, carrier parents may be completely healthy, so the family history may reveal no affected relatives, affected siblings with healthy parents, or more distantly related affected relatives. Because defective alleles are rare in most populations, the occurrence of two defective alleles (one inherited from each parent) in an individual is usually a rare event. However, if the parents are related (e.g., cousins) or from the same genetically isolated population, the likelihood of homozygosity is increased because such parents are more likely to share a recent common ancestor from whom both may have inherited the same rare allele. ^,

Where biallelic variants are detected, inference or demonstration that they substantially reduce protein function or abundance can be sufficient to infer causation. Inference of deficiency can be made by observing that the variant(s) substantially shorten the length of any translated protein (e.g., by converting an amino acid codon into a stop codon, sometimes termed a nonsense mutation); by the variant being an indel that introduces a shift in the reading frame of the transcript (a frameshift mutation); demonstration that the variant affects splicing of the transcript in a way likely to substantially affect protein function (a splicing mutation); or the variant is a SV that disrupts the gene in such a way as to prevent production of the normal protein (e.g., deleting part or all of it). Where the variant introduces a change in the predicted protein from one (or a small number) of amino acids into a different amino acid(s) without affecting the length of the protein (termed a missense mutation or small in-frame indel), additional evidence is usually needed before such a variant can be inferred to abrogate protein function. This additional evidence might include an assay demonstrating that the protein level in the blood (or other relevant tissue) of the patient is substantially reduced or if there is an established biochemical assay for the function of the protein showing this is substantially reduced (e.g., assays for the von Willebrand Factor cleavage activity of ADAMTS13 used to diagnose congenital thrombotic thrombocytopenic purpura). Sometimes, the loss of function can also be inferred (e.g., if the variant changes an evolutionary highly conserved amino acid or alters a domain that is known to be critical for protein function). Various computer algorithms that generate predictions about the deleteriousness of the variant based on such criteria exist, but such in silico evidence is obviously less reliable than a readout of protein function from a clinical assay. This is reflected in the variant annotation criteria (see “Genetic Variant Interpretation in Mendelian Diseases” later), where computer predictions only provide “supportive” evidence, whereas a clinical assay counts as “strong” evidence.

If both parents are found to be heterozygous for an allele that causes a recessive disease, the risk to each of their future offspring of having the disease is 25% since each would need to inherit a defective allele from both parents to be affected. This knowledge might have implications for parents of an affected child considering having more children. In most (but not all) recessive diseases, heterozygous relatives are at such low risk of disease that they are not offered surveillance or precluded from donating a kidney. One important exception to this is in autosomal recessive Alport syndrome, where heterozygous relatives should undergo thorough clinical appraisal and consider carefully the pros and cons of kidney donation, especially with reference to their age at the time of donation (see later).

In autosomal dominant disorders, a single defective allele can cause the disease, even if a normal allele is present on the sister chromosome. Consequently, each child of an affected individual has a 50% chance of inheriting the defective allele. This transmission risk is the same for both sexes and affected individuals often appear in successive generations. However, many autosomal-dominant disorders exhibit incomplete penetrance, meaning the probability of showing symptoms despite having a causative allele is <100%. Factors influencing penetrance include the precise mutation present in a family, environmental exposures, and genetic variation in other genes (genetic modifiers). Incomplete penetrance can make the disease appear to skip generations, and distinguishing autosomal-dominant disorders from X-linked recessive disorders can sometimes rely on identifying male-to-male transmission when eliciting the family history (making identification of such transmissions extremely valuable when taking a family history).

For an autosomal-dominant disorder to be passed to the next generation, the defective allele must allow survival to adulthood and fitness to reproduce. This could be due to incomplete penetrance or the disorder, affecting survival later in life. De novo mutations account for a significant proportion of cases, especially in potentially severe childhood-onset disorders like WT1 disorders Townes-Brocks syndrome, and Tuberous Sclerosis Complex that often occur in children without a family history. Furthermore, large, repetitive genes such as PKD1 are especially prone to such mutational events, explaining why autosomal dominant polycystic kidney disease is so common (in all human populations) and frequently occurs in individuals with no family history. Dominant disorders can also first appear in a family due to somatic mosaicism in a parent, where the mutation arises during embryogenesis and is present in the gonadal tissue but not in the affected organ, making the parent healthy but capable of passing the mutation to offspring. This implies a nonzero risk of future siblings being affected, even if both parents are unaffected.

Because they occur alongside a normal allele, the mechanisms by which dominant alleles cause disease can be more complex than simple deficiency of a protein and can be categorized into four groups:

• 1.

  Haploinsufficiency , where the presence of a single loss-of-function allele (in the presence of a normal allele on the other chromosome) results in not enough protein being made to avoid disease. This is a relatively uncommon mechanism because in most situations, transcriptional upregulation occurs, resulting in more copies of the (normal and faulty) mRNA being produced, leading to (near) normal protein abundance. Haploinsufficiency can occur when the quantity of an enzyme is critical for controlling a biological process. For instance, in PHD2 erythrocytosis, even a small decrease in the PHD2 enzyme level lowers the efficiency of degradation of hypoxia-inducible factor (HIF) 2-α, a transcription factor that promotes erythropoietin production in the kidneys, resulting in elevated hematocrit and erythrocytosis. Haploinsufficiency is also the mechanism in many disorders due to variants in genes affecting transcription factors (e.g., HNF1B), as dosage of transcription factors appears critical for many processes,

• 2.

  Gain of function, where the change in the gene, rather than reducing the function of the relevant protein leads to its increase (e.g., by impairing its normal degradation). An example of such an activating mutation is seen in HIF2-α erythrocytosis, where specific amino acid substitutions result in a protein that is more resistant to the action of PHD2 and hence oxygen-dependent degradation than is the wild-type allele, resulting in increased HIF2-α abundance, elevated hematocrit, erythrocytosis, and pulmonary hypertension. Other gain-of-function mechanisms include mistargeting mutation, such as a change in EHHADH that introduces an abnormal mitochondrial targeting motif, leading to increased abundance of this (usually peroxisomally located) protein in mitochondria, where it impairs oxidative phosphorylation in tubular cells, resulting in renal Fanconi syndrome. A third type of gain-of-function mechanism is where a change in the amino acid structure of the protein confers on the protein a new function that causes disease. An example of this is the presence of a certain single amino acid substitution in the fibrinogen protein, which confers on the protein a tendency to form amyloid fibrils that deposit in tissues, causing disease. Clearly, in such a situation, whole-gene deletion or truncating variants associated with nonsense-mediated decay would not be predicted to be pathogenic, and only highly specific missense variants (that have the required gain-of-function effect) would be expected to cause the disease. This is important to consider when trying to interpret a given variant, as the commonly used computer algorithms are designed to predict deleteriousness, but not a specific gain of function. Thus these algorithms should not be used in a disorder with a gain-of-function mechanism because that could lead to erroneous variant interpretation.

• 3.

  Dominant negative, where the variant causes a faulty protein to be produced that not only has impaired function but also inhibits the function of the wild-type protein. An example of this includes certain growth hormone receptor mutations that eliminate the intracellular domains of the protein while leaving the extracellular and transmembrane domains intact. The resulting protein is missing the segments necessary for degradation and intracellular signaling but can still bind to growth hormone (its ligand) and form dimers with the wild-type allele. Consequently, it accumulates and competes with the normal protein, interfering with growth hormone signaling and leading to congenital growth hormone insensitivity.

• 4.

  Somatic second hit, where loss of the other (functional) copy of the gene in a single cell due to a somatic mutational event renders that cell unable to make the relevant protein because it now has no functioning copy of the gene. This explains many autosomal dominant cancer syndromes, where lack of a component of cellular machinery that normally controls cell division makes the cell much more prone to uncontrolled proliferation. According to Knudson’s two-hit hypothesis, in individuals with two functional alleles of the gene in their germline genome, two separate mutational events in a cell are required to render the cell deficient in the relevant protein and hence prone to become cancerous. In those individuals who inherit one faulty copy (which is therefore present in all their cells) from a similarly affected parent, only a single somatic hit is required, making their exposure to the risk of the relevant disease orders of magnitude higher than those lacking such susceptibility. An example of this mechanism is von Hippel Lindau disease, in which loss of both copies of the VHL gene in a cell renders that the cell is unable to make the VHL protein and therefore unable to break down HIF (even in the presence of normal oxygen levels). This results in the widespread activation of a host of cellular, metabolic, and physiologic responses to hypoxia that strongly predispose the cell to developing into a clear cell renal cell carcinoma. ^, Regular monitoring to detect early renal cancers can significantly improve prognosis in such individuals, making predictive genetic testing particularly valuable for at-risk relatives of affected individuals. Diseases caused by a somatic second-hit mechanism also include nonmalignant conditions such as autosomal dominant polycystic kidney disease (where the faster accumulation of cysts in PKD1 -associated compared with PKD2 -associated disease may be attributable to the larger size, and hence greater susceptibility to mutational events, of the PKD1 gene). Somatic second-hit–mediated phenotypes may not appear until adolescence or adulthood, tend to be multifocal (arising as separate abnormalities at different times and locations within a susceptible organ), and can vary greatly in age of onset and severity within a family. This variability likely stems from the random occurrence of somatic mutations, variability in genes responsible for detecting and repairing mutations during cell division, and the influence of environmental factors (such as smoking) on the rate of somatic mutations. This latter consideration makes avoidance of carcinogen exposure during life by individuals with a somatic second-hit–mediated disease particularly advisable.

In X-linked recessive diseases, the severity or manifestations of disease are usually different in males and females. In males, the presence of a single faulty copy of a gene on the X chromosome implies the absence of a normal copy of the gene, and hence loss-of-function alleles can be inferred to be disease causing by resulting in deficiency of the protein. In females the situation is more complicated, as inactivation of one (usually at random) copy of the X chromosome in each cell means that they are essentially mosaic for the normal and faulty alleles. Whether this mosaicism has clinical consequences depends on the gene in question. In addition, it is possible that heterozygosity for a faulty X-linked gene can have haploinsufficiency or even gain-of-function effect. Therefore it is not surprising that females with either Fabry disease or X-linked Alport syndrome are at increased risk of kidney disease, although this tends to occur later in life than it does in males with these disorders.

The human mitochondrial genome has around 16,500 base pairs (compared with 3 billion for the rest of the genome) and only 37 genes, encoding only a small proportion of the components these organelles need to function, with the rest being encoded in nuclear DNA. This means that only a small proportion of mitochondrial disorders are mitochondrially inherited. Since each sperm is thought to contain fewer than 5 mitochondria (which are believed to be devoid of intact mitochondrial DNA), compared with around 200,000 within each egg, there is exclusive maternal inheritance of mitochondrial variants. Mitochondrial DNA is circular and present in multiple copies in each mitochondrion. If all the copies in all mitochondria in the egg have a variant, it will be uniformly present in all the cells of the organism (termed homoplasmy). More commonly, a genetic variant will be present in only some copies or some mitochondria (termed heteroplasmy), in which situation, as cells divide in a developing embryo, the proportion of defective mitochondrial genomes can vary across different tissues and organs. This implies that while all offspring of an affected mother are at risk of the disease, each egg contains a unique proportion of defective mitochondrial genomes that are randomly distributed in the developing embryo. Consequently, the manifestations and severity of the disease can differ significantly among affected individuals within the same family. A typical family history of a mitochondrially inherited disorder will show a wide range of severity and symptoms among family members and tellingly will never include transmission from affected males to their offspring.

Clearly, knowledge of the likely mechanism by which changes in a gene cause disease, as well as the types of genetic changes that can mediate these mechanisms, is critical for interpretation of genetic test results: Inferring pathogenicity of a deletion or early truncating mutation seen in a gene associated with disease via a gain-of-function mechanism is likely to be erroneous. While some monogenic diseases are highly penetrant, other conditions, particularly autosomal dominant diseases with onset later in life, can exhibit incomplete penetrance, with some individuals with the disease-associated genetic change(s) never becoming aware that they have the condition. In general, for autosomal dominant conditions in societies in which family size is small (typical sibships of 2–4 individuals), once the penetrance falls below 50%, the observation of a Mendelian pattern of inheritance becomes unusual and is even more so where penetrance is below 20%. For some diseases, the penetrance can be critically sensitive to the environment individuals are exposed to or to the presence of (non-Mendelian) genetic factors elsewhere in the genome, so a given genetic change may exhibit Mendelian inheritance in some, but not all, families it is present in. This illustrates that there is no clear-cut definition of what constitutes a monogenic disease allele versus a risk factor and implies that, when counseling patients, clinicians need an understanding of genetics that is broader than the classical concepts described by Mendel.

When a rare genetic variant is identified in an individual in a gene known to have been reliably associated with a Mendelian disease, in most testing systems a molecular genetics specialist, sometimes in consultation with the clinician, will endeavor to interpret the finding. Determining whether the finding can be reliably inferred to explain disease in that patient is therefore clinically actionable. As detailed earlier, the extremely consequential (and often predictive) uses of this information mandate a high degree of certainty before such variants are reportable as diagnostic findings. For clinicians involved in genetic testing, the overriding consideration is not the specificity of the test (specificity is the degree of certainty the laboratory has that a variant identified is pathogenic, as expressed by the proportion of such variants identified by the laboratory that are, in fact, disease causing), but rather it is the positive predictive value of the test (which is the probability of the patient’s disease being explained by the variant identified), which is critically dependent on the prior probability that the patient’s disease will be explained by such a molecular finding . As for any medical test, the positive predictive value is the ratio of true positives to the sum of the true and false positives identified by the test and can be calculated using the specificity and sensitivity of the test, as well as the prevalence (P) of the condition in the group of patients subjected to the test using the formula:

Because genetic tests have, at best, a sensitivity of around 85% (some patients will have deep intronic or SVs presently difficult to detect), the relationship between positive predictive value and specificity is affected by the value of P, as illustrated in Fig. 44.2 .

Positive predictive value for an 85% sensitive test according to prior probability, or prevalence of the disease in the population in which the test is applied, with each line representing a specificity between 0.90 and 0.99.

Therefore where there is a high prevalence (>60%, say) of the condition in the population being tested (e.g., polycystic kidney disease in individuals with enlarged cystic kidneys and liver cysts), even a 90% certainty that a variant causes a disease is associated with a posterior probability of such a variant causing disease in a patient of >90% and so is actionable. In contrast, where the diagnostic yield in a population is below 20% (which it might be if testing is routinely performed in children with nonsyndromic, nonfamilial developmental abnormalities of the urinary tract or steroid-resistant nephrotic syndrome), the posterior probability for a variant for which there is deemed to be a 90% chance of being disease causing is below 70% and therefore unlikely to be clinically actionable. Clearly, if the laboratory is >99% certain the variant is disease causing, such a result can be clinically actionable if the prior probability is around 20%, but below a prior probability of 5%, only variants whose pathogenicity is certain can be safely used to inform consequential clinical or other decisions. Intuitively, this can be explained by considering a specificity of 99% as equivalent to reporting one false positive in every hundred samples tested. If, among 100 samples tested there are only 5 true positives likely to be identified, then of the 6 positive results reported by the laboratory, only approximately 5 (83%) of them will be true positives. Therefore for each patient receiving such a report, there is a 1:6 chance of assigning an incorrect molecular diagnosis. For a clinician who only seldom orders a test, it may feel ethically difficult to ignore the presence of a rare or otherwise suspicious variant for which the level of certainty it is disease-causing is <99% in an individual patient. However, it is important, for the reasons outlined earlier, to consider in this situation the distribution of test results across the population of patients being tested. A strong argument can therefore be made in some circumstances for not releasing information about variants for which there is uncertainty of pathogenicity.

Of course, this can also work the other way: In a disease with a specific phenotype and precise clinical diagnosis and a single genetic etiology, even uncertain variants identified in this disease gene are likely to be causative, if found in a patient with this phenotype (see also “Tubulopathies” later). This highlights the importance of close collaboration between the clinician requesting the test and the genetic scientist carrying out the test. A genetic diagnosis is not an invariable result that can be obtained in isolation, such as by simply sending a blood sample to a laboratory with the request to identify the cause of the patient’s kidney disease. Rather, identified variants must be carefully interpreted by considering multiple aspects, a key one being the clinical phenotype and how well it fits with the identified variants. In addition, a solid understanding of the disease gene is important, as in some genes pathogenic variants typically cluster in specific regions (e.g., WT1 , exons 8 and 9 ) or even polymorphisms can be pathogenic if in trans with specific other variants (e.g., p.Arg229Gln in NPHS2 ).

This need for careful interpretation can result in conflicting assessments between laboratories. To improve and standardize this, the American College of Medical Genetics and Genomics (ACMG) and the Association for Molecular Pathology (AMP) issued guidance on how to interpret genetic variants that were well summarized in a joint ACMG/AMP consensus recommendation in 2015. They proposed 5 classifications of variants with only classes 4 and 5 (i.e., pathogenic and likely pathogenic) being clinically actionable:

• 1.

  Benign

• 2.

  Likely benign

• 3.

  Variant of uncertain significance (VUS)

• 4.

  Likely pathogenic

• 5.

  Pathogenic

In addition, they also specified the criteria on which variants could be assigned to each class, in two key figures comprising the evidence framework ( Table 44.2 ) and how the components of evidence can be integrated to assign pathogenicity class ( Fig. 44.3 ).

Evidence framework for assigning attribution of pathogenicity to variants.

BP, Benign supporting; BS, benign strong; PM , pathogenic moderate; PP, pathogenic supporting; PS, pathogenic strong; PVS, pathogenic very strong.

Although requiring refinement (particularly to address ambiguities, such as what constitutes a “well established functional study”), this has provided the international community with an invaluable framework on which to base variant interpretation, taking into account what is known or inferred (computationally or experimentally) about the effect of the variant on protein structure or function; frequency of the variant in population databases; and any previous reports of that variant in disease, taking into account that any previous reports of pathogenicity should have followed similarly robust pathogenicity assessment. It should be noted that, although not strictly defined, in the 2015 consensus statement (following a survey of the community during an ACMG open forum) the term “likely pathogenic” was proposed to mean >90% certainty of a variant being disease causing. Whether this level of certainty remains optimal given the wider usage of genomic testing in the current era remains to be determined. However, the likelihood of being pathogenic as defined in the ACMG criteria is only estimated and may reflect conservative assumptions. For this reason, retrospective studies that aim to correlate the clinical phenotype with a given genetic diagnosis are important to get a better idea of the true specificity of the test. It is also important to emphasize that Mendelian disease classification is not a clinically useful way to classify non-Mendelian genetic variants (e.g., a heterozygous deleterious variant in a gene associated with a recessive disease) because there can be a high statistical degree of certainty that a non-Mendelian genetic variant has a (sometimes small) effect on disease risk—meaning that such a variant may be known to contribute to disease pathogenicity, without it being “pathogenic” in the sense that it is the clinically actionable cause of a patient’s or family’s disease. Clinical genetics programs have established resources aimed at classifying gene-disease relationships for clinical use (e.g., the crowd-sourced Genomics England PanelApp, https://panelapp.genomicsengland.co.uk ). A U.S. National Institute of Health–sponsored international effort, called ClinGen, is under way aiming to build a central resource that defines the clinical relevance of genes and variants for use in precision medicine and research ( https://www.clinicalgenome.org ). From this it becomes obvious that a genetic diagnosis can change over time: A variant previously considered causative may be reconsidered because of the discovery of an unrealistically high allele frequency in the general population (see criterion BS1 in Table 44.2 ). Indeed, it has been estimated that up to a quarter of published “disease-causing” variants may be erroneously considered pathogenic. ^, Conversely, a variant previously considered to be of uncertain significance may be reassessed as (likely) pathogenic due to its identification in multiple other patients with the same phenotype or because of new mechanistic insights in how the variant affects protein function.

Urinary (or kidney) stones are a common medical problem with a lifetime prevalence of around 10% in some countries (see Chapter 40 for further discussion on nephrolithiasis). It is known that certain Mendelian disorders (e.g., primary hyperoxaluria, Dent disease, and Bartter syndrome, see later) can cause increased susceptibility to the development of urinary stones, and genetic testing has identified a molecular diagnosis of a Mendelian disorder in up to 10% of patients in some studies, with higher proportion seen in those with younger onset of the condition. It is also known that, even among the 90% or so of sufferers without an explaining monogenic disorder, a family history of stone disease is common (up to 37% and even higher in some studies, compared with <12% of the healthy population) and heritability (i.e., the proportion of disease risk explained by genetic factors) is estimated to be around 45% or higher. Large-scale GWASs (incorporating tens of thousands of subjects) have identified common variants at around 20 different genes that are strongly associated with the condition, but together these variants only explain around 5% of the heritability. Using whole-genome sequencing data from the UK Biobank and the 100,000 Genomes Project, single (monoallelic) truncating or predicted highly damaging rare and ultrarare variants in the gene SLC34A3 were identified in around 5% of stone formers, compared with approximately 1.5% of controls, with evidence for enrichment of these variants in stone formers highly statistically significant ( P < 4 × 10 ^–10 ). SLC34A3 encodes the sodium-dependent phosphate transport protein 2C, which is necessary for normal phosphate homeostasis, and biallelic variants in this gene are established as the cause of autosomal recessive hereditary hypophosphatemic rickets with hypercalciuria (HHRH ), so the observation of monoallelic rare, damaging variants of this gene predisposing to urinary stone formation is biologically plausible and indeed consistent with previous reports of increased urinary stone disease in parents of affected children with HHRH. However, the presence of such individually rare heterozygous genetic variants also in individuals without urinary stone disease means that identification of such a variant in an individual does not reliably predict future urinary stones, and nor do such variants exhibit Mendelian (i.e., autosomal dominant) transmission of urinary stone disease in families in which they are inherited. They do, however, contribute around 9% of the previously unexplained heritability of kidney stone disease, and the magnitude of the effect of the presence of such a variant on disease risk is similar to (and appears to be additive with) the top decile of polygenic risk score. Such monoallelic rare variants can therefore be usefully considered to be risk factors for urinary stone disease but are not clinically actionable as a monogenic diagnostic finding because they do not have a great enough effect on risk to cause Mendelian inheritance or predict future disease. This is an example where the data indicate that, while clearly having an impact on disease risk, the presence of a monoallelic rare, damaging, or even protein-truncating variant in SLC34A3 is not a diagnostic finding in the Mendelian sense, and if such a finding is reported to a patient affected by the disease, it is imperative that the magnitude of the risk effect it accounts for is also communicated in a way that the patient understands and that predictive testing in healthy relatives or to allow reproductive interventions is unlikely to be justifiable. It should also be noted that the inference of actionability is critically dependent on the availability of preventive or therapeutic interventions: If in the future a drug or other intervention becomes available that substantially reduces risk specifically in carriers of such monoallelic variants, then the case for predictive genetic testing would need to be reappraised in light of the risks and benefits of the intervention. Monogenic causes of kidney and urinary tract stone disease are considered separately below.

Alport syndrome is the second commonest Mendelian cause of kidney failure and is linked to changes in three genes: COL4A3 , COL4A4 (which are situated adjacent to each other on chromosome 2), and COL4A5 (which is situated on the X chromosome). Each gene encodes an α chain of type IV collagen (termed collagen α3[IV], α4[IV], and α5[IV,] respectively), which together form the type IV collagen trimer termed collagen α3α4α5(IV), which is first produced in the kidney after birth when podocytes switch from making the collagen α1α1α2(IV) isoform to generate the postnatal glomerular basement membrane. Biallelic (i.e., homozygous or compound heterozygous) pathogenic variants in either COL4A3 or COL4A4 (diagnostic of autosomal recessive Alport syndrome, ARAS) or a single hemizygous variant of COL4A5 (i.e., in a male patient and diagnostic of X-linked Alport syndrome, XLAS) result in failure to produce any normal collagen α3α4α5(IV). This results in hematuria (typically first observed in early childhood) with a high likelihood of subsequent proteinuria, glomerular damage, impaired kidney function, and kidney failure, which is experienced before the age of 30 in more than half of affected individuals. In addition, because collagen α3α4α5(IV) is important in the ear and eye, patients frequently experience hearing loss and some will have eye problems. Indeed, it was the combination of hereditary nephritis and deafness that allowed the definition of Alport syndrome clinically in the early 20th century. Although the distribution across the body of type IV collagen isoforms containing α5(IV) is different from those containing α3(IV) and α4(IV), it is not usually possible without specialized immunostaining of biopsy material to distinguish, in an individual male patient, whether he has the autosomal recessive or X-linked recessive form of the disease (although clearly the family history may reveal this). In addition, as mentioned earlier, females with a single faulty COL4A5 allele frequently exhibit features of the disease and are at increased risk of kidney failure, which occurs at a median age of around 65 years.

In addition to autosomal recessive and X-linked forms of the disease, it is now clear that heterozygosity for a rare, damaging COL4A3 or COL4A4 variant can be associated with a renal phenotype, especially hematuria. Where individuals undergo kidney biopsy, there is almost always thinning of the glomerular basement membrane that is visible on electron microscopy (although not all individuals with such biopsy appearances will have an identifiable change in a COL4A3/4/5 gene). In addition, some patients will exhibit proteinuria (usually later in life) and if a kidney biopsy is performed at that time, podocyte foot process effacement with or without FSGS may be seen. In some such individuals there will be progressive nephron loss and kidney failure can result. This condition has variously been termed “ARAS carrier state” or “thin basement membrane nephropathy (TBMN)” and sometimes (if there is multigenerational kidney failure in the family) “autosomal dominant Alport syndrome,” although deafness or eye involvement has rarely been reported in such individuals. Previously, the term “benign familial hematuria” was used to describe this condition, but it is now widely recognized that the word “benign” in this context is misleading since a proportion of patients will develop clinically significant kidney disease. The nomenclature for this condition (i.e., the result of heterozygosity for a single COL4A3 or COL4A4 pathogenic variant) is still under discussion, but whatever is agreed by the community, clearly the risk of clinically significant disease in such people lies somewhere between that of the general population and that of females with X-linked Alport syndrome and considerably less than ARAS or for males with XLAS. Although initial reports estimated a lifetime risk of kidney failure among COL4A3 or COL4A4 heterozygotes as high as 20%, these were based on data from patients who had undergone genetic testing to ascertain their cause of kidney disease, so they were unavoidably enriched for those with a personal or family history of more severe disease course. More recently, analysis of large-scale population sequencing data indicates that up to 0.94% of the general population have a heterozygous variant in COL4A3 or COL4A4 that is associated with hematuria, and the lifetime risk of kidney failure in such individuals is now estimated to be below 3%.

The fact that heterozygous variants in the COL4A3,4,5 genes can be seen in the general population and are associated with a relatively low risk of kidney failure illustrate the difficulties in variant interpretation and highlight again the importance of close collaboration between the clinician and geneticist. If a heterozygous (likely) pathogenic variant is identified in a patient with kidney failure, as occurred, for instance, in a substantial proportion of the “solved” cases in one prominent study of more than 3000 patients with kidney failure, how do we know that such variant is actually causative? Clinical correlation is again essential. In a patient with hematuria, proteinuria, and progressive CKD (i.e., a phenotype consistent with Alport syndrome), the variant is likely causative. But what if the patient had CAKUT with only mild proteinuria? Is the variant causative and the patient have CAKUT and a (clinically inapparent) type IV collagen gene defect? Or is it a modifier? Or just an incidental finding? This example again shows the importance of close collaboration between the clinician and the testing geneticist and that a genetic diagnosis is not a simple black-and-white result that can be done in isolation from the clinical information. For this reason, many centers have regular meetings between clinicians and the genetics laboratory to jointly discuss the relevance of any identified variants in each patient. This is analogous to regular meetings between clinicians and pathologists to discuss the changes seen in the kidney biopsy of a given patient.

For an individual female patient presenting with isolated hematuria and/or a kidney biopsy showing thinning or other abnormalities of the glomerular basement membranes without a genetic test, it is not usually possible to determine whether the cause is a fault in COL4A5 (that would indicate underlying XLAS) or relates to an autosomal COL4A3 or COL4A4 genetic variant, although a careful family history can sometimes be extremely informative. Although the clinical management and indeed prognosis might be broadly similar if the cause were a single heterozygous change in COL4A4, COL4A4, or COL4A5, if she were to have children, the risk to them would differ substantially depending on the gene: In the presence of an underlying pathogenic COL4A5 variant, her male offspring would have a 50% chance of inheriting XLAS. This information should be shared with the patient at the appropriate time to inform her decision making. Conversely, in a male patient presenting similarly, distinguishing X-linked from autosomal disease would significantly inform the risk to any (current or future) male and female offspring.

The optimal criteria for offering genetic testing in patients found to have persistent hematuria of glomerular origin remain uncertain. Important indicators of an underlying Mendelian cause include a kidney biopsy showing glomerular basement membranes that are either thin or have abnormalities characteristic of Alport syndrome; a family history of hematuria or unexplained kidney disease; or extrarenal features of Alport syndrome. The choice of whom to offer testing to will depend on the clinical utility in an individual patient and particularly the certainty with which a COL4A5 variant can be excluded clinically, as well as whether urinalysis of relatives is available. Clearly, given the range of possible outcomes from genetic testing, careful counseling is needed before performing the test and, in patients too young to engage with these issues (and therefore make an informed choice about testing) and no proteinuria, an argument can often be made in favor of deferring testing or, where appropriate, seeking a genetic diagnosis in a relative. If proteinuria has developed at the time of presentation and there are no extrarenal features or affected relatives, a kidney biopsy will usually need to be considered because the differential diagnosis would include IgA nephropathy and other inflammatory glomerulopathies for which specific treatments may be beneficial. Thus for example, it has been observed that common alleles affecting genetic risk of IgA nephropathy are enriched in healthy participants of UK Biobank with isolated hematuria (i.e., with no other clinical features of kidney or structural urinary tract disease) such that the attributable rate of undiagnosed IgA nephropathy may have reached 20% in this study cohort.

While the clinical implications of finding a pathogenic COL4A5 or biallelic COL4A3 or COL4A4 variants in a patient are usually relatively clear, the best advice to offer individuals diagnosed with a single heterozygous COL4A3 or COL4A4 variant is less well understood. It seems likely that clinical features in the patient at the time of diagnosis, as well as their own family history, will substantially influence prognostication. It is hoped that better understanding of environmental, phenotypic, and genetic predictors of outcomes in this increasingly common situation will allow this question to be better answered in the future. Alport syndrome is discussed further in Chapter 71 and Box 44.2 .

Box 44.2

Clinical Assessment and Management of Patients With Alport Syndrome

Alport syndrome is the second commonest Mendelian cause of kidney failure (after autosomal dominant polycystic kidney disease) and is associated with kidney failure and other problems as early as childhood in some patients. As ever-larger series and registries of patients with Alport syndrome have been published, it has become clear that there are strong genotype-phenotype associations in the disease, with the type of mutation (in addition to the zygosity) being strongly associated with outcomes. Among patients with autosomal recessive Alport syndrome (ARAS) or males with X-linked Alport syndrome (XLAS), the commonest type of missense mutations are substitution of glycine (which is the smallest amino acid with a side chain composed of a single hydrogen atom) with a bulkier residue. Glycine residues account for more than one-third of the amino acids in collagenous domains of each collagen protein chain and are necessary to allow tight winding of the three chains to form the triple helical collagen α3α4α5(IV) protein, so glycine substitutions result in distortion of this shape, disrupting function and leading to clinical manifestations. However, since the collagen subunits trimerize via their C-terminal noncollagenous (NC) 1 domain, variants that truncate the peptide result in a much stronger loss-of-function effect since they prevent incorporation of the entire polypeptide chain into the trimer. Clinically, such variants (and indeed some NC1 domain missense variants that have a large effect on protein structure, impairing trimer-formation) are associated with kidney failure at an earlier age than collagenous domain glycine substitutions. However, there is still significant spread and overlap in terms of age at kidney failure between these categories, so while useful as a guide to risk, care must be taken when advising patients that disease course in an individual may be very different from the average for a particular type of genetic variant. Among all truncating mutations, nonsense (as opposed to frame shifting, deletion, or splicing) mutations appear to have the worst prognosis. It is not completely understood why this is the case.

While progressive loss of renal function over time allows patients with Alport syndrome to be classified as having chronic kidney disease (CKD), there are important differences between the populations of people with Alport syndrome compared with the general (i.e., unselected) CKD population: First, patients with Alport syndrome tend to be younger (at each level of kidney function); second, renal function (even stratified for age) tends to decline much more rapidly in patients with Alport syndrome, with accelerating nephron loss especially evident when estimated glomerular filtration rate (eGFR) drops below around 30 mL/min; third, in patients with ARAS and males with XLAS, proteinuria tends to increase over time, even before eGFR begins to drop. Together, these factors contribute to a much higher risk of future kidney failure at any given level of renal impairment among people diagnosed with Alport syndrome compared with unselected CKD cohorts. This massively increased risk of kidney failure indicates significant unmet need for effective therapies that protect the kidney in Alport syndrome and also justifies a more aggressive approach to available renal protective therapies, especially early in the disease, than might be considered proportionate in the setting of more common causes of CKD.

It is becoming clear that heterozygous individuals are also at increased risk of kidney disease, but the magnitude of this risk is still uncertain: Systemic review of the case series suggests that females with XLAS have an approximately 15% lifetime risk of kidney failure (at median age of 65 years), although this methodology is susceptible to the effect of ascertainment bias. It is known that women frequently exhibit worse proteinuria and sometimes decline in renal function during pregnancy, presumably owing at least in part to the effect on their susceptible filtration barrier of the associated hyperdynamic state. Patients heterozygous for pathogenic variants in COL4A3 or COL4A4 often exhibit microscopic hematuria (with thinning of GBMs almost always present on electron microscopic examination of a kidney biopsy, with thickening/lamellations only seen occasionally and segmentally in some patients) and are known to be at increased risk of proteinuria (in which situation a kidney biopsy can often show secondary FSGS, usually with segmental foot process effacement) and even subsequent kidney failure. Although numerous case series exist showing this association and there is relatively high occurrence of heterozygous COL4A3 or COL4A4 variants detected in adults presenting with unexplained kidney failure, unlike hematuria, kidney failure rarely shows Mendelian (i.e., autosomal dominant) inheritance within families of such individuals. Large-scale genomic sequencing datasets from general populations reveal that almost 1% of some populations are heterozygous for pathogenic variants in COL4A3 or COL4A4, suggesting the lifetime risk of kidney disease is probably below 3% (contrasting with figures in the older literature as high as 20% that likely resulted from ascertainment bias owing to individuals with a personal or family history of significant kidney disease being more likely to have undergone genetic testing). Because of this complexity, there is currently controversy around the nomenclature of the disease associated with autosomal heterozygosity for variants of COL4A3 or COL4A4 . The historical term “benign familial hematuria” is now thought likely to be misleading since there clearly is an increased risk of kidney failure, and the older proposal to use the label “autosomal dominant Alport syndrome” is becoming less well justified as appreciation of the low absolute risk (with lack of Mendelian inheritance) of clinically significant kidney disease (or indeed deafness, see below) in most families together with the consideration that a diagnostic label of “Alport syndrome” will likely preclude life insurance, even if the actual risk of kidney failure in most individuals with such genetic variants is now known to be extremely low. An alternative term that has been used for many years is “thin basement membrane nephropathy,” which, although some have argued is only justified in those individuals who have undergone a kidney biopsy, has the advantage of being applicable to patients with relevant clinical and biopsy features in whom genetic testing is either not available or does not reveal a (likely) pathogenic change in a relevant gene. Clearly, it is important to recognize that the term “thin basement membrane nephropathy” should not be inferred to automatically imply a benign prognosis, and long-term surveillance (to detect any hypertension and proteinuria, at least into late middle-age) is indicated in any individual with this condition.

Clinical trials using the angiotensin-converting enzyme inhibitor ramipril have provided evidence that early institution of renin angiotensin system blockade can delay kidney failure in males with XLAS. This recapitulates evidence from mouse models and fits with a paradigm in which increasing pressure across the glomerular filtration barrier accelerates nephron loss in the disease owing to the defective extracellular matrix composition in the GBM, so it is unsurprising that targeting this process pharmacologically provides clinical benefit. Clinically, the data have been interpreted by professional organizations (such as the European Renal Association), leading to recommendations that all patients with any evidence of microalbuminuria or proteinuria in the setting of X-linked or autosomal recessive Alport syndrome, or a single heterozygous COL4A3 or COL4A4 variant, should be offered this treatment. In addition, newer guidelines have advocated institution of renin angiotensin system blockade from the age of 2 years if hematuria is evident, even in the absence of any proteinuria, in hemizygous males with XLAS or any patient with ARAS. There is currently less strong evidence for the use of sodium glucose transporter 2 inhibition in Alport syndrome, but the magnitude of the benefit in terms of reduction in progression to kidney failure among adult patients with proteinuric chronic kidney disease (and the fact that patients with Alport syndrome were not excluded from large-scale trials of this therapy ^, ) has provided justification for its use in clinical practice, with a retrospective case series suggesting that it is well tolerated in this group. It is hoped that trials of this class of medication in children with kidney disease will provide direct evidence that allows quantification of safety and efficacy in Alport syndrome.

As a multisystem disorder affecting hearing and the eyes, as well as the kidney, XLAS and ARAS can have significant implications for affected children and young people and guidelines recommend annual childhood screening for hearing impairment from the age of 4 years in boys diagnosed with XLAS and all children with ARAS, since language and educational development can be impacted if hearing loss is unrecognized in childhood. Frequent surveillance for visual impairment is not mandated given existing childhood screening programs for myopia in many countries, but an awareness of the risk of keratoconus, which may manifest as severe myopia, is important for patients and parents of affected children. Eye and ear complications are rare in people with a single autosomal pathogenic variant, and although hearing loss is commonly seen in older individuals, alternative causes (genetic or acquired) should be sought in young people in this category with significant hearing impairment.

If kidney failure occurs, the best current treatment modality is kidney transplantation, but since many patients will have affected (or heterozygous) first-degree relatives, access to living-related kidney donors is reduced among people with Alport syndrome compared with non-Mendelian causes of kidney disease. An important question relates to the suitability of heterozygous relatives (for instance, mothers of boys with XLAS) to donate a kidney. Guidelines suggest that other options should be explored first, and the pros and cons of predonation kidney biopsy should be discussed with such donors in order to identify subclinical evidence of kidney damage. While there is no empiric evidence that allows quantification of the effect on risk of future kidney disease in heterozygotes donating a kidney, it is generally assumed that it is greater than for people of similar health, age, and sex with normal type IV collagen genes, so decisions should be made taking into account both this unquantified risk and also the potential benefit to the recipient. Data revealing whether allografts from heterozygotes fare less well are also lacking, further affecting the quality of advice that can be shared with patients in these discussions. Having said that, kidney donation by heterozygous females who might become pregnant in the future is generally discouraged since the combined risk factors of their genotype and having a single kidney during pregnancy are difficult to quantify.

A rare complication of kidney transplantation in patients with Alport syndrome is de novo production of an antibody response directed against type IV collagen of the allograft glomerular basement membrane. With current immunosuppression regimens, this risk is now estimated to be below 3% and far lower in patients with at least one missense variant. While immunohistochemistry can show linear staining of the glomerular basement membranes, serologic tests for antiglomerular basement membrane antibodies are not thought to be sensitive in this situation unless the biallelic missing chain is collagen a3(IV), which contains the Goodpasture antigen, which is detected serologically.

Only gold members can continue reading. Log In or Register to continue

Share this:

• Click to share on Twitter (Opens in new window)
• Click to share on Facebook (Opens in new window)

Related posts:

Why Are Disorders of Sodium Concentration So Difficult? Preparing Dialysis Patients for Renal Transplantation Psychosocial Issues in Renal Transplantation Kidney Drainage and Percutaneous Tract Closure Disorders of ECF Volume: Congestive Heart Failure Approaches to the Induction of Tolerance

Stay updated, free articles. Join our Telegram channel

Join