Review Of Genetic Testing in Kidney Disease Patients: Diagnostic Yield Of Single Nucleotide Variants And Copy Number Variations Evaluated Across And Within Kidney Phenotype Groups Ⅱ

4 | DISCUSSION 

Determining true diagnostic yield from the papers studied is difficult because of the variability in patient, cohort, and test characteristics. The diagnostic yield for the different phenotype groups should therefore be thought of in terms of ranges, and recommendations should not be based on single studies. Given the enormous variation in key parameters, we wanted to avoid overinterpretation and therefore did not perform statistical analyses. However, the comprehensive overview presented will help weigh the possible relevant factors with potentially useful information for clinical practice. This overview allows clinicians to judge which studies are most relevant for their specific patients/patient groups and estimate an a priori probability of finding a genetic cause for their patients.

FIGURE 3 Scatterplots and box plots showing the relation between diagnostic yield and cohort characteristics. (a) Scatterplots show the relationship between diagnostic yield and the number of cases sequenced within a specific study. Legend describes a number of studies for which data on this cohort characteristic was available. Colors indicate the disease group from which studies were derived. Each dot represents for one study what the number of sequenced cases was in that cohort and what diagnostic yield was obtained from that same study. (b) Boxplots representing the diagnostic yield in clinical cohorts versus research cohorts. The lower half shows clinical versus research cohorts across the different disease groups. Legend describes a number of studies for which data on this cohort characteristic was available.

Not surprisingly, we found that diagnostic yield was higher based on expected patient characteristics (e.g., family history, consanguinity, extrarenal features, and young age of onset) within studies. When we assessed these same characteristics plotted against the diagnostic yield between the different studies (instead of within one study), this pattern was not seen for consanguinity and young age of onset. This might be explained by a combination of other characteristics having a larger impact on the diagnostic yield. Also, within the adult-onset group, there is a large variation in disease severity (e.g., ESKD at age 20 has a higher diagnostic yield than ESKD at age 70) and the likelihood of finding a monogenic cause (i.e., in a typical ADPKD cohort a high diagnostic yield is expected (Figure 2e)). We did find an indication that young age of onset is related to a higher diagnostic yield based only on CNVs (Figure 2f). This is likely explained by genome-wide CNV analysis being performed more often in this group (Supplementary Table 3).

CLICK HERE TO GET TO KNOW A HERBAL FORMULATION OF CISTANCHE FOR KIDNEY AND SEXUAL FUNCTION 

This review includes an overview of CNV yield per phenotype. CNVs that are not picked up by regular sequencing can be assessed using a separate test (e.g., SNP-array) or by a CNV calling tool based on sequencing data (Knoers et al., 2022). The different types of tests that were used in the included studies varied greatly, as did the number of genes that were analyzed (i.e., covering one gene, the requested multigene panel or exome/genome-wide and the impact this has on the diagnostic yield [Supplementary Table 3, Supplementary Figure 3]). We found the highest contribution of CNVs to diagnostic yield in CAKUT, ciliopathies, and tubulopathies. However, across all phenotype groups where CNV testing was performed, CNVs did contribute to the diagnostic yield and CNV analysis should be considered when genetic testing is performed. The high contribution of CNVs to diagnostic yield in CAKUT patients confirms previous reports (Knoers et al., 2022). CNVs were extensively investigated in CAKUT patients as SNV yield is relatively low in this group. It is yet to be established for some other phenotypes whether the diagnostic yield based on CNVs is underestimated hitherto as we found that in many phenotype groups CNVs were not investigated (Figure 1c).

Surprisingly, a higher number of genes tested did not always correlate with a higher yield. In theory, this would always be the case in comparable cohorts. However, the cohorts we studied vary distinctly. On the one hand, we describe cohorts with a highly likely monogenic cause (such as ADPKD-suspected patients) requiring only a small number of tested genes to result in a high diagnostic yield. On the other hand, we find a lower diagnostic yield and an increase in the number of tested genes in less highly suspected cohorts. Cohort size did not appear to correlate with a number of tested genes (data not shown). Another explanation for this finding is an increase in the number of tested genes in patients where testing of common known disease genes did not result in a genetic diagnosis. Since it is possible that a proportion of these unsolved cases have either a genetic diagnosis in a not yet discovered gene, or a non-monogenic cause explaining their disease, a lower yield in this group can be hypothesized. Also, in some cohorts, patients with known mutations were excluded, but the total number of these patients with a mutation was not reported (Bekheirnia et al., 2017; Braun et al., 2016; Faure et al., 2016; Heidet et al., 2017; Kohl et al., 2014; Schueler et al., 2016; Vivante et al., 2017; Ziyadov et al., 2021). Finally, in some studies, genetic testing in specific known genes was not performed (e.g., known CAKUT genes (Caruana et al., 2015; Sanna-Cherchi et al., 2012)). For translation of the reported diagnostic yield to clinical practice, it is important to know these details. It was beyond the scope of this review to analyze whether for all cohorts the panel composition included all known causal genes, including appropriate phenocopy genes, for each phenotype at the time that that specific study was performed.

FIGURE 4 Scatterplots and box plots showing the relation between diagnostic yield and test characteristics. (a) Scatterplots showing the relationship between diagnostic yield and the number of genes sequenced within a specific study. Legend describes a number of studies for which data on this specific test characteristic was available. Not all studies were included since in some studies the number of genes sequenced differed within the study. Colors indicate the disease group from which the study was derived. Each dot represents for one study what the number of sequenced cases was in that cohort and what diagnostic yield was obtained from that same study. (b) Pie charts visualizing a number of studies that performed either single nucleotide variant (SNV) or copy number variation (CNV) testing or both. (c) Boxplots representing the diagnostic yield in relation to the type of variants that were tested. In the right panel, this is split into the different disease groups.

We found that the diagnostic yield decreased in larger cohorts. This might be explained by smaller cohorts being more clearly defined and therefore having a higher suspicion of a monogenic cause. In addition to selection bias within a study, publication bias might also explain this finding. The larger cohorts probably give a more reliable estimation of the true diagnostic yield in a relatively unselected kidney disease population. We would like to point out that a published cohort with a high diagnostic yield is not necessarily the “better” cohort to use for clinical decision-making. When applying tightly restricted criteria genetic diagnoses are likely missed. When interpreting published cohorts for clinical practice, one should also take into account that there are many local differences in clinical phenotyping and diagnosing, especially when seen from an international perspective. For example, a patient with unknown CKD in one center might differ distinctly from a patient with unknown CKD in another center.

We were curious to find out if we would find a higher yield in research cohorts or diagnostic cohorts, although the distinction between these two was not always clearly described. One could assume that selection bias is higher in research cohorts resulting therefore in a higher yield. However, research cohorts are often gathered over the course of multiple years and phenotyping is not always on point. When interpreting publications on research cohorts, one should take into account that access to clinical information and variant interpretation, including opportunities for variant segregation, might differ from a clinical diagnostic setting and are likely less individualized. Clinical cohorts might be more reliable for obtaining a diagnostic yield that is generalizable to a diagnostic setting. However, clinicians might also miss cases that should be offered genetic testing. Furthermore, information on the originating population from which a cohort was derived from can be lacking in clinical cohorts, especially when the clinical cohort describes the diagnostic yield in a genetic center. We found that the diagnostic yield between the two types of cohorts did not differ distinctly. For studies from either cohort type, it remains unknown whether there is an overrepresentation or an underrepresentation of genetic cases. Performing a study that actually tests all kidney disease patients in a clinical setting would answer this question

FIGURE 5 Top 50% of diagnostic yield explained by a limited number of genes and/or copy number variations (CNVs). (a) Pie charts visualize a number of genes responsible for the top 50% of yield. The legend on the left describes for each category in the pie chart, the number of studies identified with this number of genes. In between brackets on the right are the number of studies that reported on the top 50% causal genes and separated by “j” the number of genes responsible for the top 50% for that specific phenotype. (b) The genes that made up this  50% in each study per phenotype group are displayed here unless genes were responsible for only one positive case and/or multiple genes made up for the final percentages. * Variants in the large variable number tandem repeats region of MUC1 are usually missed by massively parallel sequencing

To derive a minimum diagnostic yield for a specific diagnostic setting, we extrapolated the diagnostic yield to the larger cohort from which the tested (clinical) cohort was derived. An interesting example of this is presented by the studies of Snoek et al. and Schrezenmeier et al., who both report a diagnostic yield of 21% in a kidney transplant (waitlisted) cohort (Schrezenmeier et al., 2021; Snoek et al., 2022).

5 | IMPACT OF GENETIC DIAGNOSIS 

A genetic diagnosis can have a diagnostic, prognostic, and treatment impact. The studies we selected for the review highlight this (Table 3). Multiple studies reported on the molecular genetic diagnosis resulting in correction of the clinical diagnosis. While percentages vary between these different studies, all highlight the potential importance of establishing the correct diagnosis through genetic testing. The therapeutic impact varies from referral and evaluation for previously unrecognized extrarenal features to changing treatment plans. A clear example of the therapeutic consequences of genetic testing is SRNS; most genetic forms of SRNS do not respond to immunosuppressive drugs and can therefore be spared the potential toxicity of these ineffective medications. A clear example of prognostic impact is an extremely low disease recurrence in many genetic kidney diseases following kidney transplantation as opposed to kidney diseases with a nongenetic cause.

Importantly, identifying a genetic cause can be crucial for the patient and/or parents of a patient with regard to genetic counseling; it informs recurrence risks and can support patients' and parents' decision-making regarding reproductive options such as prenatal and pre-implantation genetic diagnosis. In addition, family members can be counseled about disease risk, presymptomatic testing, and screening options for secondary signs in first-degree family members whilst not performing genetic testing. A genetic diagnosis can also be of importance for living kidney donation by family members. What impact a genetic diagnosis has, including therapeutic impact, will depend on the phenotype, but also on the individual circumstances of a patient and family, and the local/regional availability of treatment and family planning options.

FIGURE 6 Summary of key findings from literature review (n = 115 articles). The left side of the figure summarizes the characteristics influencing diagnostic yield and the impact of a genetic diagnosis. The right side of this figure summarizes the additional key take home messages. *Estimated by the authors of this review that this is due to tightly restricted phenotype criteria.

6 | CONSIDERATIONS FOR GENETIC TESTING

It is important to note that the type of test that is chosen for genetic testing in a patient can have a big impact on the chance of finding a genetic cause. The technological advancements in genetic testing approaches have made (MPS-based) CNV testing and exome-based sequencing possible and the advantages are being recognized over the years (Supplementary Figure 2). Gene panel composition, number of tested genes, and CNV analysis can influence the likelihood of finding a genetic cause. In addition, in ADTKD-suspected cases, it is important to consider MUC1 testing or additional PKD1 testing in ADPKD suspected cases after MPS-based multigene panel or exome testing. In only 6 studies (including 3/4 ADTKD cohorts) additional testing was performed to detect a cytosine insertion in the variable number tandem repeats region of MUC1 that is usually missed by MPS (Supplementary Table 3) (Kirby et al., 2013). Nineteen studies reported additional tests to reach sufficient coverage of all PKD1 exons, which is challenging because of the existence of six pseudogenes (PKD1P1-6) with 97.7% sequence identity. Fifteen of these studies focused on ciliopathy phenotypes and four included mixed kidney disease phenotypes. One study reported the exclusion of ADPKD patients because PKD1 is not well-captured by WES (Lata et al., 2018). Clinicians that request genetic testing need to be aware of what disease-causing variants in what genes can be detected by what genetic test and when additional genetic tests should be requested (Knoers et al., 2022; Köttgen et al., 2022).

Sometimes clinicians who do not yet have expertise in (nephron) genetics seem to assume that “performing WES” will cover any disease-causing variant. One aspect of this false assumption is CNV detection. While WES-based CNV analysis is upcoming, this is not yet always included. It is therefore good to consider whether, for a specific patient, additional CNV testing is needed. In the studies included in this review, many different tools were used, with differences in yield per tool being reported (Moreno-Cabrera et al., 2020; Yao et al., 2017). Some studies covered CNVs in all genes, while others focused only on genes within a gene panel (Supplementary Figure 3). Also, separate tests (e.g., MLPA covering one gene or genome-wide SNP array) were performed for CNV detection. Other aspects of this false assumption are variations in gene coverage, difficult-to-sequence regions (e.g., repeat regions, pseudogenes), and noncoding variants.

In the past, gene panels were always enrichment-based, meaning that only the set of genes that were selected prior to sequencing would be sequenced and analyzed. Today, diagnostic labs are often using exome-based gene panels (Supplementary Figure 2b). With this approach, the complete exome is sequenced, but only the genes of interest from a specific gene panel are analyzed. An advantage of WES and the usage of exome-based gene panels is the efficient method in which data is derived including the possibility to analyze additional genes without having to resequence the patient's DNA (Knoers et al., 2022). WES also makes it possible to reanalyze or identify phenocopies. The studies included in this review highlight this. Warejko et al. identified phenocopies in 4% of patients in an SRNS cohort (Warejko et al., 2018). Also in the NS cohort of Landini et al., reverse phenotyping of patients let to the diagnosis of phenocopies in 28% of cases (Landini et al., 2020). Szab
o et al. found phenocopies in 22% of patients with ARPKD (Szab
o et al., 2018). In a cohort with various phenotypes, Riedhammer et al. discovered that 19% of diagnosed cases were a phenocopy (Riedhammer et al., 2020). Phenocopies and local differences in clinical phenotyping are arguments for a broader gene panel composition. Broad genetic testing also has significant challenges, including a higher chance of incidental findings, and the difficulties in interpretation of variants of unknown significance (VUS) (Bertier, Hétu, & Joly, 2016). This should be taken into account when choosing an initial smaller gene panel or a broad multigene panel or an exome-wide analysis. Also, counselors should be comfortable with, and skilled in counseling these findings and when to refer to a clinical geneticist. Availability of different genetic tests, genetic care and agreement on what test can be requested by nongeneticists will differ per country.

In all the phenotypes described in this review, only a limited number of genes are responsible for the top 50% of established diagnoses, highlighting the relevance of core genes for phenotypes (Martin et al., 2019). Even though only a limited number of genes are responsible for the top 50% of genetic diagnoses, the study of Groopman et al. found that 39/66 detected monogenic disorders were detected in only a single patient (Groopman et al., 2019). In this same study, four genes were responsible for 54% of confirmed diagnoses. Rao et al. reported that 15 genes accounted for 61% of genetic diagnoses, but in total, 106 distinct monogenic disorders were detected in a cohort of 1,001 pediatric patients with clinical suspicion of genetic kidney disease (Rao et al., 2019). Therefore, we recommend considering the analysis of a complete set of known kidney disease genes after a first negative focused exome-based panel result, also in light of the possibility of phenocopies. We recommend not starting with this large set of genes to avoid unnecessary variants of unknown significance and incidental findings. This recommendation is again dependent on the availability and costs of reanalysis per country. While this review focuses primarily on diagnostic yield across and within kidney phenotype groups, additional genetic testing considerations for kidney disease patients can be found in recently published recommendations (Knoers et al., 2022; Köttgen et al., 2022).

7 | STRENGTHS AND LIMITATIONS 

Our systematic approach to selection and data extraction resulted in the most extensive overview of diagnostic yield in nephrogenesis yet. Given the challenges in comparing the different studies, we only summarized and visualized the data but did not perform a statistical meta-analysis. One important variable was variant classification which was not identical in all studies. In 59/115 articles only the American College of Medical Genetics and Genomics (ACMG) criteria were used and in 4/115 ACMG criteria were used together with other filtering steps (Supplementary Table 3). Some studies that used other variant classifications than the ACMG criteria did use the same variant descriptions (i.e., pathogenic and likely pathogenic variants) as shown in the pivot table in Supplementary Table 3. We did find that most recently published papers used ACMG criteria (Supplementary Figure 2d). Benson et al. highlights the impact that different classification criteria can have on the reported yield by reporting a diagnostic yield of 68% using ACMG criteria and a yield of 81% using Mayo Clinic pathogenicity guidelines. Furthermore, it was not always reported whether both likely pathogenic and pathogenic variants were included. In some cases, VUS were included in the reported diagnostic yield and it was not possible to subtract these, which might give an unjustly high yield. With WES-based panels and ACMG criteria being used more often, comparing future studies might be easier. We chose to include studies published in the last 10 years, but even with this limitation, we expect that the diagnostic yield was likely higher in the more recent articles because of an increase in the number of known kidney disease genes and novel techniques. Although our data do not clearly support this notion (data not shown), this is likely explained by other factors masking this effect. Future reporting on diagnostic yield would also benefit from reporting of the population of which the study population was derived from. This was not always reported in the included articles in this review, rendering interpretation of the minimum diagnostic yield in these (clinical) cohorts challenging

8 | WHAT IS NEXT? 

Although this review gives an extensive overview of reported diagnostic yield, there are still gaps in knowledge regarding the true prevalence of hereditary kidney disease in the general CKD population. To get an estimate of this, a study would need to be set up in which all kidney disease patients seen in different clinical settings would receive genetic testing. Since this might not be feasible, we would like this to be at least a call for action to report in detail on the population reported in diagnostic yield studies, to report on the patient, cohort, and test characteristics mentioned in this review, and to use standard variant classification protocols. Also, opportunities for the segregation of variants (of unknown significance) and CNV analysis should not be missed. Another opportunity for diagnosing additional patients is WGS, which makes it possible to sequence the entire genome. For ultimately determining the true diagnostic yield there is still a long road ahead since there are likely still many coding and noncoding genetic causes involved in kidney disease that need to be discovered.

9 | CONCLUSION

This review gives an overview of the diagnostic yield of genetic testing across and within kidney disease phenotypes. The most important findings and key take home messages are summarized in Figure 6. We confirm that patient characteristics (e.g., family history, consanguinity, extrarenal features, and young age of onset) can positively impact the diagnostic yield. Furthermore, we emphasize the impact of the specific genetic test requested, including its ability to reveal CNVs. We also show the importance of considering the kind of cohort in which a study was performed, for interpreting the reported yield. We show that a genetic diagnosis can have a diagnostic, therapeutic, and prognostic impact. Considering reclassifications based on genetic findings and the possibility to obviate the need for a diagnostic renal biopsy in many cases, a genetics-first approach can be considered in clinical practice for establishing the patient's diagnosis. The number of genes to examine, whether and how to perform CNV analysis and, in ADTKD/ADPKD additional tests to cover for MUC1 and PKD1 need to be weighed when requesting a genetic test. Of course, it is important to note that patient and family-specific situations can also influence the decision to do a genetic test, and also what genetic test is chosen. In addition, the availability of genetic testing in different countries can have an impact on the accessibility of genetic testing. This review gives clinicians guidance on estimating an a priori probability of finding a genetic cause for kidney disease in their patients.

AUTHOR CONTRIBUTIONS

Rozemarijn Snoek, Laura R. Claus, and Albertien M. van Eerde set up the design for the review. Nine V. A. M. Knoers provided structural feedback on the study design and progress. Rozemarijn Snoek extracted data from a subset of articles to define the variables of interest. Laura R. Claus extracted, evaluated, and analyzed the data and drafted the paper. Albertien M. van Eerde and Nine V. A. M. Knoers critically assessed the paper. All authors approved the final version of the manuscript. ACKNOWLEDGMENTS The authors acknowledge and thank Rieko Haring and Richard van Kemenade who ran the literature database searches and screened articles for eligibility as part of their bachelor thesis. This work was supported by the Dutch Kidney Foundation (18OKG19 to A. M. v. E.). The authors of this publication are members of the European Reference Network for Rare Kidney Diseases (ERKNet). CONFLICT OF INTEREST The authors declare no conflict of interest.

DATA AVAILABILITY STATEMENT Data sharing is not applicable to this article as no new data were created or analyzed in this study.

REFERENCES 

Adalat, S., Hayes, W. N., Bryant, W. A., Booth, J., Woolf, A. S., Kleta, R., … Bockenhauer, D. (2019). HNF1B mutations are associated with a Gitelman‐like tubulopathy that develops during childhood. Kidney International Reports, 4, 1304–1311. Ahn, Y. H., Lee, C., Kim, N. K. D., Park, E., Kang, H. G., Ha, I. S., … Cheong, H. I. (2020).

 Targeted exome sequencing provided a comprehensive genetic diagnosis of congenital anomalies of the kidney and urinary tract. Journal of Clinical Medicine, 9. Al Alawi, I., Al Salmi, I., Al Rahbi, F., Al Riyami, M., Al Kalbani, N., Al Ghaithi, B., … Sayer, J. A. (2019).

 Molecular genetic diagnosis of Omani patients with inherited cystic kidney disease. Kidney International Reports, 4, 1751–1759. Al Alawi, I., Molinari, E., Al Salmi, I., Al Rahbi, F., Al Mawali, A., & Sayer, J. A. (2020). Clinical and genetic characteristics of autosomal recessive polycystic kidney disease in Oman. BMC Nephrology, 21, 1–11. Al‐Hamed, M. H., Al‐Sabban, E., Al‐Mojalli, H., Al‐Harbi, N., Faqeih, E., Al Shaya, H., … Meyer, B. F. (2013)  A molecular genetic analysis of childhood nephrotic syndrome in a cohort of Saudi Arabian families. Journal of Human Genetics, 58, 480–489. Al‐Hamed, M. H., Kurdi, W., Alsahan, N., Alabdullah, Z., Abudraz, R., Tulbah, M., … Albaqumi, M. (2016). The genetic spectrum of Saudi Arabian patients with antenatal cystic kidney disease and ciliopathy phenotypes using a targeted renal gene panel. Journal of Medical Genetics, 53, 338–347.

Supportive Service:

Email:wallence.suen@wecistanche.com 

Whatsapp/Tel:+86 15292862950

Shop:

https://www.xjcistanche.com/cistanche-shop