Exome sequencing (ES) of a pediatric cohort with chronic endocrine diseases: a single-center study (within the framework of the TRANSLATE-NAMSE project)

Background Endocrine disorders are heterogeneous and include a significant number of rare monogenic diseases. Methods We performed exome sequencing (ES) in 106 children recruited from a single center within the TRANSLATE‑NAMSE project. They were categorized into subgroups: proportionate short stature (PSS), disproportionate short stature (DSS), hypopituitarism (H), differences in sexual development (DSD), syndromic diseases (SD) and others. Results The overall diagnostic yield was 34.9% (n = 37/106), including 5 patients with variants in candidate genes, which have contributed to collaborations to identify gene-disease associations. The diagnostic yield varied significantly between subgroups: PSS: 16.6% (1/6); DSS: 18.8% (3/16); H: 17.1% (6/35); DSD: 37.5% (3/8); SD: 66.6% (22/33); others: 25% (2/8). Confirmed diagnoses included 75% ultrarare diseases. Three patients harbored more than one disease-causing variant, resulting in dual diagnoses. Conclusions ES is an effective tool for genetic diagnosis in pediatric patients with complex endocrine diseases. An accurate phenotypic description, including comprehensive endocrine diagnostics, as well as the evaluation of variants in multidisciplinary case conferences involving geneticists, are necessary for personalized diagnostic care. Here, we illustrate the broad spectrum of genetic endocrinopathies that have led to the initiation of specific treatment, surveillance, and family counseling. Graphical Abstract


Background
Endocrine disorders comprise a wide spectrum of different diseases (affecting the hypothalamus and pituitary, thyroid gland, adrenal cortex, sexual differentiation and growth disorders, polyendocrine and neoplastic disorders) [1].Autoimmune disorders, environmental factors or medications can lead to endocrine dysfunction, but genetic etiology contributes significantly to the spectrum of pediatric endocrine disorders.
In recent years, the use of next-generation sequencing, such as exome sequencing (ES), has increased significantly in almost every medical specialty.However, to the best of our knowledge, there are few data on genetic diagnoses by ES in a representative endocrine cohort from a single center.
To improve the care of patients with rare diseases, the German Federal Joint Committee (G-BA) funded the innovation project TRANSLATE-NAMSE from April 2017 to September 2020 [2].
TRANSLATE-NAMSE was a healthcare project to establish new structures and processes across different healthcare providers and disciplines.Ten German centers for rare diseases (Berlin, Bonn, Bochum, Dresden, Essen, Hamburg, Heidelberg, Luebeck, Munich, Tuebingen), two health insurance companies (AOK Nordost; Barmer GEK) and the Alliance for Chronic Rare Diseases (ACHSE e.V.) formed a consortium to design, test and evaluate a model of structured care for patients with rare diseases [2].Here, in a retrospective single-center study, we elucidated the spectrum of genetic variations underlying rare endocrine diseases.We describe in detail the clinical and genetic findings of 106 children and adolescents with endocrine disorders who underwent ES.

Study design
Children and adolescents with various endocrine diseases were recruited consecutively from December 2017 to February 2020 within the framework of the healthcare project TRANSLATE-NAMSE in the endocrinological outpatient clinic of the University Children´s Hospital in Heidelberg [2].The analysis of the results of ES was conducted under the guidelines of the Ethics Committee of the University of Heidelberg (S-690/2020) and in accordance with the current version of the Declaration of Helsinki (2013).Initially, baseline and dynamic endocrine testing were performed, possibly supplemented by imaging procedures.Inclusion criteria were a) an endocrine disorder or symptom complex with at least one endocrine disorder, b) no established genetic diagnosis and c) no alternative causal explanation for the condition, such as an autoimmune disease.These patients were divided into subgroups: proportionate short stature (PSS), disproportionate short stature (DSS), hypopituitarism (H), differences in sexual development (DSD), syndromic diseases (SD), and others.For detailed clinical characteristics, see Table 1.Most exome analyses were performed as triosequencing (parent-child-trios; n = 72, 67.9%) followed by single-analyses (n = 26, 24.5%).Duo-exomes were performed in 5 patients (4.7%), and quattro-exomes were performed in 3 patients (2.8%).Phenotypes were compiled using Human Phenotype Ontology (HPO) terms [3].
ES was performed in the laboratories of the Institute of Human Genetics, Technical University Munich, the Institute of Human Genetics, Tuebingen and the Institute of Human Genetics, Heidelberg.All patients received information and care at Heidelberg University Children's Hospital.The indication and subsequent evaluation of this specific diagnostic tool was made in multidisciplinary case conferences, always including pediatric endocrinologists and geneticists.All patients or their legal guardians gave written consent for both the genetic diagnostics and the TRANSLATE-NAMSE project with retrospective analysis and publication of the data.Unsolved cases with variants in candidate genes were reanalyzed in multidisciplinary case conferences after 2 years.

Exome sequencing
ES was performed in Munich using the SureSelect Human All Exon Kit (Agilent, 60 Mb V6) for enrichment and a NovaSeq6000 (Illumina, San Diego, CA, USA) for pairedend sequencing.Reads were aligned to the Human Genome Assembly GRCh37 (hg19).Allele frequency estimation was performed using in-house databases and the Genome Aggregation Database (gnomAD).Variants were analyzed with a MAF < 1% (autosomal-recessive inheritance) and >0.01%for de novo variants.In addition, a phenotypebased search was conducted by performing an OMIM full term search using the three most characteristic phenotypic traits to establish a gene list.The filter queries variants with a MAF < 0.1%.Moreover, CNVs with a MAF < 0.01 and mtDNA variants with a MAF < 1% were assessed.Identified variants were classified according to the American College of Medical Genetics and Genomics (ACMG) guidelines [4][5][6].In Tuebingen, diagnostic ES and data analysis were performed according to a quality-controlled standard operating procedure essentially as described previously [7].In brief, coding genomic regions were enriched using a SureSelect XT Human All Exon Kit V.7 (Agilent Technologies, Santa Clara, California, USA) for subsequent sequencing as 2×100 bp paired-end reads on a NovaSeq6000 system (Illumina, San Diego, California, USA).Generated sequences were analyzed using the megSAP pipeline (https://github.com/imgag/megSAP),and prioritized genomic variation (SNPs, indels, CNVs, and SVs) was classified with reference to the ACMG guidelines.One ES was performed at the German Cancer Research Center (DKFZ), Heidelberg, Germany, on DNA of the affected girl and both parents as previously described [8].All results were evaluated in a multidisciplinary case conference to match the clinical and genetic findings.Only patients with likely pathogenic or pathogenic variants according to ACMG (hereafter referred to as "diseasecausing") in established disease genes were included as solved cases in the overall diagnostic yield.Five individuals with variants in candidate genes subsequently established as disease genes were also categorized as solved and assigned to the overall diagnostic yield.Individuals with (1) negative results (i.e., no variant[s] prioritized), (2) variants of uncertain significance (VUS) in endocrinopathy-associated genes, or (3) variants that did not explain the phenotype were summarized as unsolved patients.The nomenclature of DNA sequence variants was controlled using VariantValidator [9].

Statistics
Statistical analyses were performed using SPSS Statistics 28.0.1.0(IBM, Armonk, New York).The results are presented as the mean with standard deviation and median with range.Comparison of the diagnostic yield was performed using Fisher's exact test and Pearson's chi-squared test.

Patient characteristics
We included a total of 106 patients with a mean age of 9.6 ± 6.5 years (range: 0.1-28.1 years, median: 9.9) at the time of initiation of exome diagnostics during 2017-2020.Sixty-eight were male (64.2%), and 38 were female (35.8%).For clinical characterization, see Table 1 and Supplementary Tables 1 and 2.

Genetic findings
Exome sequencing initially identified disease-causing variants in 32/106 individuals, representing a diagnostic yield of 30.2%.Five patients had six variants in genes that had not been associated with monogenic disorders at the time of the initial analysis.However, based on international collaborations and subsequent re-evaluation in multidisciplinary conferences, a probable causal association has been established (Table 2).Reanalysis of unsolved patients at 2 years revealed 40 monogenic diagnoses in 37/106 patients due to Mean age ± SD (years) 6.6 ± 7.1

Table 2
List of patients in which the variants could be assigned to be disease-associated (n  3 patients with dual diagnoses.The overall diagnostic yield was 34.9%.Thirty/40 diagnoses (75.0%) were ultrarare diseases with a prevalence < 1 in 50.000 [10].

Novel gene-disease associations
In five individuals, variants were prioritized in genes that were not associated with monogenic disorders at the time of data interpretation.In the majority of these individuals (n = 4), de novo variants were found.Only one patient had compound heterozygous variants in the candidate gene MCM7.All variants were absent in gnomAD [11,12].All potential novel gene-disease associations were discussed by multidisciplinary teams and submitted to GeneMatcher [13,14].Three individuals were subsequently published within large collaborations linked by GeneMatcher [15][16][17].A total of five novel gene-disease associations have been identified.Respective variants were found in PPP1R12A [15], SMARCA5 [16], H4C3 [18], MCM7, and BDNF.Separate manuscripts are in preparation for these genes.As previously indicated, these five patients were considered solved and included in the overall yield.For patient characteristics, see Table 1 in the Supplementary Appendix.

Secondary findings
In patient 43, a microdeletion of Xq27.1 affecting the F9 gene was reported as a secondary finding, and the diagnosis of hemophilia was clinically confirmed (Table 3 and Supplementary Table 2).In patient 44, a homozygous missense variant (class 5), previously reported multiple times to be pathogenic, was reported as a secondary finding, as JAK  [19].The patient was subsequently clinically evaluated but showed no clinical signs of Aicardi-Goutieres syndrome, and neopterin in CSF was normal, as was the interferon signature.As the variant was identified in a healthy additional adult individual in a homozygous state in the in-house exome database, the multidisciplinary case conference concluded a reduced penetrance of RNASEH2B-associated Aicardi-Goutieres syndrome.
Sibling patients 38 and 39 with hypopituitarism harbored a likely benign variant in the HESX1 gene, which was inherited from the unaffected mother.In patients 40, 41, 42 likely pathogenic variants or VUS were identified and reported back to the multidisciplinary teams.After intensive rephenotyping and discussion in case conferences, they were classified as "not disease-causing" (Table 3 and Supplementary Table 2).

Dual diagnoses
Interestingly, we identified three patients with more than one genetic diagnosis.Patient 29 had a disease-causing variant in  WRN, which explains the multiple features of accelerated aging in addition to microcephaly mental retardation.A homozygous variant of uncertain significance was identified in ASPM, but the multidisciplinary board considered the diagnosis of primary autosomal recessive microcephaly 5 to be likely.Patient 36 had congenital primary hypothyroidism.In addition to a disease-causing variant in DUOX2, we also identified a homozygous splice variant in DUOX1.Given the unusual severe presentation, the multidisciplinary board classified the variant as a likely disease modifier.Patient 37 suffered from neurofibromatosis type 1 and PHEX-related hypophosphatemia.Patient characteristics are shown in Supplementary Table 2.

Discussion
As more genetic alterations are discovered, particularly in ultrarare diseases, and next-generation sequencing becomes more widely available, we wanted to evaluate whether ES is a useful diagnostic tool in children and adolescents with endocrine disorders.
In an unselected cross-sectional cohort of 106 patients from a single endocrine center, ES identified likely pathogenic/pathogenic variants in 34.9% of previously undiagnosed patients with endocrine disorders.Of the confirmed diagnoses, 75% were ultrarare diseases.Here, we extend the list of disease-associated genes in 5 cases.An accurate phenotypic description including comprehensive endocrine diagnostics as well as the evaluation of variants in multidisciplinary case conferences involving geneticists are necessary for personalized diagnostic care and initiation of specific treatment, surveillance, and family counseling.
A genetic diagnosis was established in 37 patients, including three patients with dual diagnosis, and causative variants were found in 40 genes.Most patients (n = 22) with a suspected underlying endocrine disorder were immediately substituted with appropriate hormones after pathological endocrine testing.After confirmation of the genetic diagnosis, these families were offered genetic counseling and were generally advised to continue hormone treatment.In one girl, hypophosphatemic rickets was confirmed as an unexpected secondary diagnosis, and treatment with burosumab was initiated.In three other patients, growth hormone treatment was started after the genetic diagnosis of Noonan syndrome and risk assessment together with the parents.
In certain syndromes, such as Noonan or Werner syndromes, individualized lifelong surveillance is recommended.In this cohort, extended surveillance was indicated in at least 7 patients (6.5%) after genetic diagnosis.
The majority of diagnoses were based on de novo variants or autosomal recessive inheritance (both 40%).The percentage of autosomal recessive disorders in our cohort was surprisingly high compared to previous studies in patients with neurodevelopmental disorders, which showed a low contribution (4-16%) of autosomal recessive disorders [26,29].In patients with endocrine disorders, the mode of inheritance has not been systematically investigated [22,26].
The significant proportion of de novo variants highlights the utility of trio sequencing as a first-line strategy, especially in sporadic cases.Even in highly recognizable syndromes or defined endocrinopathies due to specific hormonal constellations, we discovered unexpected diagnoses.For example, patients with pathogenic variants in NFKB2 or IGSF1 or the dual diagnosis of patient 37 (neurofibromatosis and hypophosphatemic rickets) would have been missed by targeted panel sequencing.An important finding of this study is the high prevalence of ultrarare diseases (75.0%).In addition, we extended the list of disease-associated genes associated with endocrinopathies to facilitate variant classification in other patients.Three results have been published [15][16][17], and the others are in preparation for manuscript, highlighting once again the potential of international data sharing and collaboration [13,14].We strongly recommend interdisciplinary case conferences involving pediatric endocrinologists and geneticists to discuss patient phenotypes and known constellations together with genetic findings to classify new variants.Although controversial, some authors suggest the diagnostic yield as a parameter of effectiveness [27].We believe that an individualized decision in a multidisciplinary case conference is preferable to avoid repetitive single gene or panel testing, as the cost of NGS techniques has decreased and this approach avoids the diagnostic odyssey of patients with complex rare diseases, thus anticipating personalized patient care and the most appropriate treatment.One limitation is that these results are derived from a single center.Therefore, the data should be confirmed in large international studies to evaluate the overall diagnostic yield in children and adolescents with endocrine disorders.

Conclusions
The present study shows that ES is an effective tool for genetic diagnostics in pediatric patients with complex endocrine disorders.An accurate phenotypic description, including comprehensive endocrine diagnostics by pediatric endocrinologists together with variant evaluation by geneticists in multidisciplinary case conferences, is necessary for specific results and personalized clinical management.Furthermore, we were to expand the list of disease-associated genes in 5 cases.For the first time, we estimated the diagnostic yield in different groups of endocrinopathies.It was highest in complex patients.Finally, the broad spectrum of genetic endocrinopathies, including ultrarare diseases in 75% of patients, was demonstrated, leading to the initiation of specific treatment, surveillance, and family counseling.
A > G, p.(Gln106Arg); c.350 T > G, p.(Leu117Arg) C > T, p.(Ala167Val), c.946 C > T, p.(Arg316Cys) could be assigned to be disease-associated were reevaluated in multidisciplinary conferences in regard of phenotype, laboratory findings, and family history.b At time of data interpretation unknown.

Fig. 1
Fig. 1 Variant classification of 106 patients with endocrine disorders.The bars show the distribution of variants according to the diagnoses.VUS variant of unknown significance, DSD disorder of sexual development; n = patients

Fig. 2
Fig. 2 Distribution of all disease-causing variants based on the mode of inheritance, with de novo variants representing the most common mode of inheritance

Table 3
List of patients with variants not related to the phenotype* (N = 5) and secondary findings ɫ (N = 2) *indicates patients with variants not related to the phenotype ɫ indicates secondary findings