Prenatal diagnosis of HNF1B-associated renal cysts: Need to differentiate intragenic variants from 17q12 microdeletion syndrome?

Objective Large deletions of chromosome 17q12 (17q12DS) or intragenic variants in HNF1B are associated with variable developmental, endocrine and renal anomalies, often already noted prenatally as hyperechogenic/cystic kidneys. Here, we describe pre- and postnatal phenotypes of seven individuals with HNF1B aberrations and compare their clinical and genetic data to previous studies. Methods Prenatal sequencing and postnatal chromosomal microarray analysis was performed in seven individuals with renal and/or neurodevelopmental phenotypes. We evaluated HNF1B-related clinical features from 82 studies and reclassified 192 reported intragenic HNF1B variants. Results In a prenatal case, we identified a novel in-frame deletion p.(Gly239del) within the HNF1B DNA binding domain, a mutational hotspot as demonstrated by spatial clustering analysis and high computational prediction scores. The six postnatally diagnosed individuals harbored 17q12 microdeletions of varying size. Literature screening revealed highly variable reporting of HNF1B-associated clinical traits. Overall, developmental delay was more frequent in 17q12DS carriers, although both mutation groups showed a high phenotypic heterogeneity. The reclassification of all previously reported intragenic HNF1B variants provided an up-to-date overview of the mutational spectrum. Conclusions We highlight the value of prenatal HNF1B screening in renal developmental diseases. Standardized clinical reporting and systematic classification of HNF1B variants is necessary for a more accurate risk quantification of pre- and postnatal clinical features, improving genetic counseling and prenatal decision-making.


INTRODUCTION
The HNF1B (hepatocyte nuclear factor-1-beta, MIM *189907; also known as transcription factor-2 (TCF2)) protein is a transcription factor which, together with its dimerization partner HNF1A, belongs to the homeodomain-containing superfamily.
[7] SNVs/indels described to date include nonsense, frameshift, splice-site and missense variants. The latter are mainly distributed within exons 2-4, which code for the 19,22,27] All described HNF1B whole gene deletions so far, correspond to chromosome 17q12 microdeletions spanning 1,3 to 1,8 Mb on average and including neighboring genes. [10,28,29] Both, intragenic variants and genomic rearrangements contribute almost equally to the HNF1B-associated phenotype. [11] Importantly, numerous studies reported neurodevelopmental disorders (NDDs) in individuals with 17q12 microdeletions (MIM #614527) containing HNF1B. [10,[28][29][30][31][32][33] Although NDDs were considered an exclusive feature of chromosome 17q12 microdeletion syndrome (17q12DS), there are recent reports also in individuals with intragenic HNF1B alterations. [16,[34][35][36] In this study we report on a series of seven individuals with intragenic HNF1B aberrations or 17q12DS and describe their renal and extra-renal phenotype prenatally and in childhood or adult life. By comparing this cohort to reports from current literature, we discuss the potential need to differentiate intragenic variants from 17q12DS in the increasingly common setting of prenatally diagnosed kidney abnormalities. This question is not only important for prenatal decision-making but also for postnatal management in affected individuals.

Individual 1 (I1)
A pregnant woman presented to our Department of Obstetrics and Gynecology at gestational week 21 5/7 for sonographic examination. An isolated bilateral renal cortical hyperechogenicity without kidney enlargement was noted in the fetus. Amniotic fluid volume was normal, and amniocentesis was performed. Prenatal ultrasound at 32 weeks of gestation additionally revealed bilateral fetal pyelectasis. The female individual I1 was born at gestational week 36 6/7 with normal birth parameters. During the neonatal period she manifested a temporary respiratory adjustment disorder and feeding problems. Postnatal kidney ultrasound at 3 months of age demonstrated a regression of renal hyperechogenicity, bilateral renal cysts of varying size, with the largest being 3 mm, an enlarged right kidney with a total volume of 24 ml (97. P), renal pelvis dilatation grade I (8 mm) of the left kidney and bilateral megaureters of 2 mm. Sonographic examination of liver and pancreas exhibited no structural anomalies. Blood glucose level was slightly increased (110mg/dl), and urine albumin/creatinine ratio was 504mg/g. Plasma magnesium and uric acid concentration as well as liver transaminases were normal. Additional postnatal findings included a minor atrial septal defect. Ocular anomalies were not detected. At the time of last assessment at age of 3 months, development and growth were in the normal range. Renal sonographic examination of both parents did not reveal any abnormalities (Fig. 1C, 1D, Tab. 1 and Supplementary note).

Individuals 2-7 (I2-I7)
Clinical data of individuals I2-I7, who presented with renal and/or neurodevelopmental phenotypes at our genetic outpatient clinic and at the Center for Rare Diseases Erlangen, were retrospectively collected for this study (Fig. 1B)

Computational analyses of HNF1B variants
The clustering analysis of described HNF1B variants, the collection of these variants from the literature together with computational analysis of all possible single amino acid deletions and missense substitutions as well as the protein structure analysis of the Gly239del variant are described in the Supplementary note and all data is provided in Supplementary File 1. In brief, we collected an up-to-date list of 217 intragenic HNF1B variants from 88 articles and public databases, harmonized these according to the HGVS nomenclature, annotated them with different computational scores and evaluated their pathogenicity according to the ACMG criteria. To analyze the spatial clustering, we plotted the distribution of these variants across the linear protein representation and estimated p-values from empiric distributions of drawing the observed number of missense variants in the respective domains. Additionally, we used the published tertiary protein structure of HNF1B to analyze the potential effect and proximity to other pathogenic variants of the herein identified c.715_717del, p.(Gly239del) variant.

Prenatal identification of a pathogenic in-frame deletion in individual I1
Prenatal panel sequencing in individual I1 revealed a novel heterozygous 3 base pair (bp) deletion in exon 3 of HNF1B (NM_000458.3: c.715_717del), resulting in an in-frame deletion of the highly conserved amino acid (AA) glycine at position 239 (p.(Gly239del)) (Fig. 2B). The identified indel variant was not present in the Genome Aggregation Database (gnomAD) and was computationally predicted (CADD, MutationTaster) to be deleterious. A potential splice effect was not detected (Human Splicing Finder, version 3.1). The c.715_717del variant could not be identified in DNA from peripheral blood lymphocytes of both parents and thus likely arose de novo. Structural modeling based on the crystal structure of the HNF1B protein showed that the glycine at position 239 (Gly239) lies in an alpha-helix motif of the homeodomain, which has DNA binding function. The deletion of Gly is predicted to disrupt two neighboring AAs (Trp238 and Lys237) involved in binding to DNA (Fig. 3 were reported in a boy with bilateral renal cysts, kidney failure and diabetes in childhood and in a girl with MODY, renal cysts in right kidney, agenesis of left kidney and pancreas atrophy, respectively. Furthermore, a variant affecting the neighboring residue Trp238 (c.712T>C, p.(Trp238Arg)), was described as pathogenic (Fig. 3).
[9] Based on the ACMG criteria[42] (de novo occurrence, no inclusion in population databases, previously described variant at the same position, affected highly conserved AA, computationally indicated pathogenicity), we classified the identified variant as pathogenic (class 5).

17q12 microdeletions in individuals I2-I7
Postnatal high-resolution CMA analysis in individuals I2-I7 revealed 17q12 microdeletions of variable sizes ranging from 1.42 Mb in I5 to 1.80 Mb in I6 ( Fig. 2A). De novo occurrence was confirmed in all cases apart from individual I3, where the deletion was excluded in the mother, but paternal DNA was not available (see also Tab. 1 and Supplementary note).

Clinical features of the affected individuals
Prenatal renal anomalies such as hyperechogenic kidneys were only identified in individual I1 with the in-frame deletion in HNF1B (Fig. 1C). Postnatally, 4/7 individuals exhibited renal cysts. Individual I1 was diagnosed with multiple renal cysts in early infancy (Fig. 1D), I3 and I5 in puberty and I6 in adulthood. Other renal abnormalities included kidney enlargement, renal pelvis dilation and bilateral megaureter in individual I1 as well as right kidney aplasia and left kidney hypoplasia in individual I7. Recurrent urinary tract or kidney infections from childhood on were reported in individuals I6 and I7. Individuals I5 and I7 developed stage II and end-stage renal disease in their late teens and I6 end-stage renal disease after the age of 26. In individuals I2 and I4 no renal manifestations were reported at last follow-up, at the age of 8 years and 8 months, and 12 years, respectively.
The most prominent extra-renal phenotypes in individuals with 17q12DS were NDDs and behavioral problems (Fig. 1B). Individual I2 exhibited learning difficulties, I3 mild intellectual disability and Asperger syndrome and I4 a low-normal IQ with learning and concentration difficulties. Behavioral problems varied from oppositional behavior (I2) and hyperactivity and restlessness (I3) to autistic-like behavior (I4). Motor milestones were delayed in I4 and I6.
Individual I6 had a single seizure attack in adulthood. Finally, MRI showed enlarged brain ventricles in individual I4 (Tab. 1 and Supplementary note).

Reviewed intragenic variants and clustering in protein domains
We retrieved a total of 217 intragenic HNF1B variants from databases, 88 published articles and this report (I1) (Supplementary File 1). Only 46/216 variants (21.3%; excluding the variant in individual I1) were deposited in public databases as of 2018-12-02. After manual review, 192 of them could be classified as ACMG class 4 (likely pathogenic; n=57) or 5 (pathogenic; n=135). For four variants described in the literature there was enough evidence to classify them as ACMG class 2 (likely benign), while for 21 variants only insufficient information was available, and they were classified as ACMG class 3 (uncertain significance). The majority of (likely) pathogenic variants in HNF1B are truncating (59.4%; and 75 (c.410_484del) bp are annotated as in-frame variants deleting 10 (p.(Ala373_Gln383delinsGlu)) or 25 (p.(Arg137_Lys161del)) AAs, respectively. While these two variants are annotated as "disruptive" (Fig. 2B, marked in purple below protein scheme) because they ablate multiple AAs which likely disrupts the protein structure, the herein identified in-frame deletion of glycine at position 239 (p.(Gly239del)) is annotated as "conservative" (Fig. 2B, marked in purple above protein scheme). Besides affecting an AA position previously described as mutated in patients with HNF1B-associated disease, the deletion is located within the HNF1B homeodomain, which mediates DNA binding.
The HNF1B homeodomain and the second half of the N-terminal domain ("HNF-1_N") host the two mutational hotspots for missense variants, while likely gene disruptive variants are dispersed throughout the protein (Fig. 2B, middle panel). The possible missense variants in these two domains additionally have significantly higher CADD scores when compared to those of the C-terminal domain ("HNF-1B_C") or non-domain regions of the HNF1B protein ( Fig. 2B, lower panel and Fig. S1).

DISCUSSION
Improved sonographic technologies increasingly enable the identification of fetal hyperechogenic and/or cystic kidneys. Nevertheless, identifying the underlying cause and specifying prognosis remain challenging. The recurrent 17q12 microdeletion encompassing HNF1B constitutes a frequent differential diagnosis for fetal renal anomalies with variable postnatal consequences ranging from isolated renal to more complex, syndromic manifestations. [11,15,19,22] The emerging role of HNF1B intragenic variants for prenatally detected renal and extra-renal phenotypes has only recently been recognized. [11] Panel sequencing of common genetic causes for prenatal-onset kidney anomalies in individual I1 with hyperechogenic kidneys in fetal ultrasound, identified the novel in-frame single AA deletion c.715_717del, p.(Gly239del) in HNF1B. This is the first report of an inframe HNF1B variant leading to a renal phenotype resembling that of previously described HNF1B cases. One of the main questions in this prenatal situation was further prognosis.
Due to the variable presentation of HNF1B aberrations with renal and extra-renal plus possible neurodevelopmental aspects, this is particularly challenging. Currently prenatal prognosis and decision-making regarding TOP depend mainly on clinical presentation including oligohydramnios and detectable congenital anomalies. However, other issues to consider during prenatal genetic counseling without severe complications are frequency and severity of the various postnatal manifestations.
Although very young, individual I1 with the intragenic HNF1B variant did not show any signs of developmental delay. Clissold and colleagues provided evidence for association of NDDs exclusively with 17q12 microdeletions but not with HNF1B intragenic variants.
[30] The predominant hypothesis for the neurodevelopmental phenotypes was haploinsufficiency of other genes encompassed by the deletion. [28,29,31,33,43] Until recently, only single cases of NDDs in carriers of intragenic HNF1B variants had been reported. [16,34,36] However, a recent study described NDDs in (6/54) 11% of individuals with MODY harboring HNF1B intragenic alterations; this frequency was significantly higher when compared with cases who had another cause of diabetes and not significantly different from 17q12DS cases (17.0%; 9/53). Postnatally, the presence of accompanying organ anomalies is important to guide appropriate management and surveillance. In the present study 5/7 (71.4%) individuals with either HFN1B intragenic variant or larger deletion had renal manifestations, in agreement with the previously reported 61-91% prevalence in HNF1B cases (Tab. 2).[9, 48] Absence of renal disease in individuals I2 and I4 may be attributed to a milder renal phenotype overshadowed by developmental deficits [33] or to their young age. Even though renal anomalies were only prenatally reported in individual I1 in our cohort, we cannot exclude that they were present but not detected in the remaining cases with renal manifestations identified after birth (I3 and I5-I7). In the literature, the severity of prenatal renal phenotypes is not correlated with the type of HNF1B aberrations.
[11] Notably, HNF1B intragenic variants progressively result in more severe renal function impairment and poorer renal prognosis postnatally than 17q12 deletions.[9, 30] Accordingly, renal cysts were diagnosed in individual I1 directly after birth, whereas 4/6 (66,7%) 17q12DS individuals showed renal abnormalities later in infancy, puberty or early adult life. Regardless of the initial clinical indication for testing, identification of 17q12 deletions should always be accompanied by regular blood and urine tests, and sonographic kidney evaluation.
The frequency of diabetes in individuals with HNF1B-related clinical phenotypes is highly variable, ranging from 9-82% with a mean-age of diagnosis at 24 years, but also possible neonatal onset (Tab. 2). [9,12,48,49] Consistently, individual I1 showed slightly increased levels of blood glucose in early infancy, whereas two 17q12DS individuals (I5 and I6) were diagnosed in puberty. The remaining individuals did not exhibit diabetes to date, yet at least for individuals I2-I4 this may relate to their young age. The severity of diabetes was shown to be similar in both carriers of truncating and missense variants. Dubois-Laforgue et al.
reported a lower BMI and more severe diabetes observed at diagnosis in 17q12DS as compared to intragenic variant cases.
[9] In addition to annual HbA1c and glucose level measurements both, carriers of 17q12DS and intragenic HNF1B variants, should be trained to self-monitor diabetes symptoms in order to enable early diagnosis and avoid complications.
Other noteworthy clinical features of the herein described individuals are short stature in individual I4, uterus aplasia in individual I7 and strabism in individuals I4, I5 and I7.
Gonadotropin treatment increased the predicted final adult height of individual I4 for about 10 cm. Yet, reports on growth restriction in individuals with 17q12DS [29] or HNF1B intragenic variants [36] are rare, and the efficacy of growth hormone treatment in these cases needs elucidation. Uterus aplasia is within the HNF1B-associated spectrum, with intragenic variants and whole deletions reported in 18-50% of women with uterine and renal anomalies. [9,24] Thus, obstetric evaluation already at young age is required. Finally, detailed ophthalmic examination is warranted, since in 40-44% of HNF1B-aberration carriers strabism and other ocular anomalies were reported (Tab. 2). [10] The reasons for the phenotypic variability within the two groups of individuals with HNF1B intragenic alterations and whole deletions, as well as between these groups, currently remain unclear. The clinical diversity observed within the 17q12DS carriers could be attributed to the incomplete penetrance and variable expressivity of this CNV, similar to other recurrent microdeletions.
[50] A quantitative impact of the 17q12 microdeletion with a relatively small size effect, which is difficult to be ascertained from the small cohorts examined, could for instance justify the presence or absence of NDDs in affected individuals. [50,51] Regarding clinical differences within the intragenic variants group, different pathomechanisms, especially for the missense variants, constitute a possible hypothesis. Indeed, a dominantnegative effect, allowing a residual function of the HNF1B protein, rather than haploinsufficiency has been discussed for some variants.   [8,19,27,56] or in postnatally diagnosed children with renal diseases [11,15,22,30,34,36], HNF1B mutation screening should be an integral part of prenatal diagnosis for hyperechogenic/multicystic kidneys. Customized targeted panels capturing multiple cystic kidney disease-related genes, like the one used herein, have several favorable features for the prenatal setting: high coverage, short turn-around times and compatibility with small benchtop sequencing machines. Clinical exome sequencing (covers ~5.000 to 7.000 disease genes), is currently a reasonable compromise, although it requires more time and is less accessible. Regardless, NGS should be recommended concurrently with CMA for rapid prenatal diagnosis in fetuses with hyperechogenic or cystic kidneys. However, the application of next-generation methods will consequently result in the increased identification of pathogenic variants and variants of unknown significance (VUS). To facilitate fast and effective assessments, all identified variants should be systematically classified and deposited in appropriate databases, as performed in this study for all described in the literature so far (Supplementary note and Supplementary File 1).
In conclusion, similar to previous studies our study indicated no differences in the prenatal renal phenotypical spectrum between intragenic HNF1B aberrations and 17q12DS, and a wide range of postnatal renal and extra-renal abnormalities for both. The most prominent postnatal phenotypic differences, potentially affecting prenatal decision-making, are a more severe progress of renal impairment in individuals with intragenic HNF1B variants, and the increased NDD risk, which is a confirmed feature in 17q12DS and suspected in HNF1B intragenic variant carriers. Incomplete references to the overall clinical presentation of the affected individuals in the published reports and previous classifications of identified variants not following current recommendations of a standardized 5-tier system add to the difficulty of risk estimation. We therefore systematically collected HNF1B-related clinical traits, revised and harmonized all intragenic variants from 88 articles and propose a minimal reporting scheme for HNF1B-associated disease (Tab. 2). Building on these efforts, future studies using a standardized ascertainment of phenotype/genotype information will improve our understanding and characterization of HNF1B-associated disorders, and enable the optimization of genetic counseling.

Funding information
A.R. was supported by the German Ministry of Education and Research (BMBF, grant numbers: 01GS08160, 01GM1520A (Chromatin-Net)) and the IZKF Erlangen (E16).

Ethical aspects
Informed written consent was obtained from all patients or their legal guardians, and the study was approved by the Ethical review board of the Friedrich-Alexander-Universität Erlangen-Nürnberg.

Conflicts of interest
The authors declare that they have no conflict of interest.

Supplementary notes
Detailed descriptions for cases I2-I7 and additional methods, results and figures.

Supplementary File 2
Excel file containing the worksheets "summary", "CP02-Panel_Genes", "CP02-Panel_Design", "HNF1B_primers", "PKD1_LR-PCR" and "PKD1_Seq" describing the targeted panel design and primers used for additional sequencing. The "summary" worksheet contains a detailed description of all other worksheets and the respective data columns.  N  o  v  e  l  g  e  n  e  s  f  o  r  a  u  t  i  s  m  i  m  p  l  i  c  a  t  e  b  o  t  h  e  x  c  i  t  a  t  o  r  y  a  n  d  i  n  h  i  b  i  t  o  r  y  c  e  l  l  l  i  n  e  a  g  e  s  i  n  r  i  s  k   .  2  0  1  8  ,  b  i  o  R  x  i  v  .  4  7  .  D  e  c  i  p  h  e  r  i  n  g  D  e  v  e  l  o  p  m  e  n  t  a  l  D  i  s  o  r  d  e  r  s  ,  S  .  ,   P  r  e  v  a  l  e  n  c  e  a  n  d  a  r  c  h  i  t  e  c  t  u  r  e  o  f  d  e  n  o  v  o  m  u  t  a  t  i  o  n  s  i  n  d  e  v  e  l  o  p  m  e  n  t  a  l  d  i  s  o  r  d  e  r