Introduction

Hydrocephalus (HC) is characterized by aberrant cerebrospinal fluid (CSF) dynamics (with or without ventricular dilation) that can lead to increased intracranial pressure. When left untreated, HC may be fatal and cause severe impairment in neurodevelopment. While classical theories of CSF posited that CSF is produced predominantly in the lateral ventricles via the choroid plexus and flows through the foramina of Monroe, third ventricle, cerebral aqueduct, and the fourth ventricle where it is disseminated through the central canal of the spinal cord and the subarachnoid space to be reabsorbed by arachnoid granulations [1], this model is no longer considered dogmatic [2]. Anatomical disruption of CSF flow and/or CSF pulsatility may result in a buildup of CSF to be classified as obstructive or non-communicating HC. However, HC can be communicating (i.e., no obvious anatomical blockade of absorption or obstruction), the result from increased production of CSF in response to injury, impaired absorption from the subarachnoid space, or result from defects in cortical development. These insults, in turn, may lead to ventricular dilation, among other potential and highly debated pathophysiologic mechanisms. Importantly, the global burden of HC is high [3], with significant morbidity and mortality regardless of treatment [4]. However, the genetic and mechanistic basis of HC remains poorly understood, largely due to the genetic complexity and phenotypic heterogeneity of the disease as well as cost of large-scale human genetics studies.

HC is a component of a wide-array of genetic syndromes [5], a secondary consequence of brain injury (intraventricular hemorrhage (IVH), infection, etc.) [6, 7], and a component of many central nervous system congenital abnormalities (i.e., neural tube defects, Chiari malformation, etc.) with a number of comorbid phenotypes including epilepsy and autism, among others. HC is a highly polygenic disease [8,9,10], with genes of varying functions and mechanisms conferring risk to the disease. The current treatments for HC are surgical interventions such as insertion of a ventricular (-peritoneal, atrial, etc.) shunt or endoscopic third ventriculostomy (ETV), which may be combined with choroid plexus cauterization (CPC) [6, 11]. While many studies have evaluated the efficacy and cost of these procedures [12], long-term morbidity of HC remains high and both treatments are prone to failure [13, 14]. Furthermore, while clinical trials have attempted pharmacological strategies to treat HC [15], no pharmacological treatment has been successful. In addition, HC may present in adulthood as normal pressure hydrocephalus (NPH). A more sophisticated and detailed understanding the genetic architecture and molecular pathogenesis of HC may lead to development of targeted pharmacologic treatments.

While numerous studies have aimed to identify causative genetic mechanisms leading to HC, largely based on isolated human case studies and murine models [5], critical limitations include cost, patient/family recruitment, number of patients (small by population-genetics’ standards), individual variant validation (typically de novo mutations), and very important species differences between model-organisms and human disease. Proposed pathophysiological mechanisms of HC include impaired development of the neural stem cell niche [16,17,18,19,20], abnormal ciliated ependymal cells [21,22,23], disruption of the ventricular zone [24, 25], and primary alterations in CSF absorption and/or secretion [26,27,28,29]. However, our understanding of these mechanisms is derived from varied model systems, which do not always accurately recapitulate the genetic and pathophysiological basis of human HC. Furthermore, there is increasing evidence that germline genetic variation contributes to risk of HC [5, 8, 10, 30]; however, most cases of HC remain genetically undefined and clinical genetic testing is rarely performed.

Elucidation of the genetic architecture of both shared and etiology-specific forms of HC may uncover pathophysiological mechanisms and correlate genetic risk factors with clinical and surgical outcomes, with the potential to directly influence surgical counseling and clinical management. While many genes have been implicated in the pathogenesis of HC in humans, the study designs, approaches, and levels of evidence identifying and validating these genetic findings vary greatly. Uncovering the genetic basis of HC relies on many factors, but most importantly on the clinical phenotype in question because HC rarely occurs in isolation. Comorbid phenotypes (neural tube defects, primary structural brain disorders, epilepsy, cognitive delay, etc.), and antecedent injuries – IVH and/or infection (meningitis, intracranial abscess, and/or sepsis), alone and in concert, confound most classical approaches to understanding genetic disease. Advancing our understanding of HC genetics, therefore, will necessitate understanding the extent to which co-occurring phenotypes are present and integration of multiple molecular and genetic data. Furthermore, elucidation of human-specific molecular mechanisms necessitates study in human tissue representative of the diverse populations HC affects. Here we summarize genetic studies of HC in humans and offer suggestions for advancing the field forward.

Methods

Search criteria

The US National Library of Medicine PubMed database and the Online Mendelian Inheritance in Man (OMIM) were queried for English-language studies using Title/Abstract, MeSH headings, key words, and genetic descriptors relevant to genetic causes of HC and ventriculomegaly. The OMIM database was used as an additional adjunct database as well. Our search terms are included below. Duplicates identified across multiple databases were identified. We strictly adhered to PRISMA guidelines [31].

Our PubMed search syntax included the following: (HC[Title/Abstract]) OR (Ventriculomegaly[Title/Abstract]); ((HC[MeSH Major Topic]) OR (Ventriculomegaly[MeSH Major Topic])) AND ("mendelian" OR "de novo" OR "functional genomics" OR "whole exome sequencing" OR "whole-genome sequencing" OR "genotyping" OR "genotype" OR "microarray" OR "genome-wide association study" OR "genome wide association study" OR "GWAS" OR "transcriptome wide association study" OR "transcriptome-wide association study" OR "TWAS" OR "gene expression" OR "copy number variation" OR "insertion" OR "deletion" OR "mosaic" OR "mosaicism" OR "genetic variation" OR "consanguineous" OR "consanguinity" OR "autosomal recessive" OR "autosomal dominant" OR "x-linked recessive" OR "x-linked dominant" OR "inherited" OR "inheritance" OR "non-coding" OR "coding" OR "co-expression" OR "germline" OR "linkage" OR "linkage disequilibrium" OR "genetic counseling" OR "syndrome" OR "syndromic" OR "genetic testing" OR "aqueductal stenosis" OR "obstructive HC" OR "acquired HC" OR "congenital HC" OR "proteomics" OR "proteomic" OR "metabolomic" OR "metabolomics" OR "methylation" OR "mutation" OR "genetic deficiency" OR "gain of function" OR "gain-of-function" OR "loss of function" OR "loss-of-function" OR "molecular"[Title/Abstract]).

We next queried the Online Mendelian Inheritance in Man (OMIM) database [32] using the search terms: “HC” or “ventriculomegaly” to identify genetic disease of which HC is a component. The search returned 671 entries which were manually reviewed. Duplicates within the OMIM database were excluded (n = 95). The resulting search query resulted in 3,709 studies.

Inclusion and exclusion criteria

Records (n = 3,709) from the above search were initially evaluated via abstract and screened for exclusion criteria: (1) Records published before 1970; (2) no genetic data of any kind; (3) no HC diagnosis; or (4) animal subjects. A total of 2,652 studies were excluded. Full text screening of the remaining papers (n = 1,057) was then screened for inclusion criteria. The second round of screening was carried out by full text review (n = 1,057). The same exclusion criteria were applied, while inclusion criteria were implemented: (1) Records published after 1970; (2) pediatric cohort (0–18 years of age); (3) primary genetic analysis; (4) confirmed diagnosis of human HC; and (5) human subjects. The final records were assessed for eligibility and records unavailable in English were excluded (n = 2). The final studies included (n = 327) were then evaluated for the methodology and type of genetic analysis performed. The papers included in our study (n = 327) were then subject to secondary analyses to assess for (1) change in number of publications over time; (2) geographic and ethnic associations of HC; (3) size of study; (4) central nervous system and non-central nervous system phenotypic associations. Figures were created using BioRender.

Author affiliation and subject country of origin

Authors’ institutional affiliation was obtained via PubMed’s “Affiliations” tab within the respective research articles PubMed webpage. Authors were then cross-referenced via Google search to increase validity of institutional affiliation at the time the study was performed. Articles were individually queried for the country of origin of patients with HC. If not explicitly stated, it was assumed that the patients were from the same country as the senior author’s affiliation. The total number of cases were tallied and tabulated on a world map using OpenStreetMap.

Results

HC in humans can be caused by or is secondary to several factors including structural brain disorders, cilia abnormalities, brain tumors resulting in CSF obstruction requiring CSF diversion, neural tube defects, prematurity and germinal matrix fragility, neonatal systemic and CNS infections, intracranial hemorrhage, evolutionary selection pressures, and ‘genetic’ anomalies, classically thought as Mendelian disorders (Fig. 1). Thus, we conducted a systematic review of human genetic studies of HC to quantify and summarize the current state of genetic contributions to HC of various etiologies (Fig. 2). However, genetic susceptibility confers risk to all these preceding factors as well as to HC directly. Thus, understanding the pleiotropic effect of genes on both risk factors and development of HC is needed and requires highly detailed phenomics analysis [33]. Here, we summarize all genetic studies of human HC across the age spectrum, including discussion of animal models of HC only as corroborating findings of genes and pathways identified in humans where there is a reasonable degree of evolutionary conservation. We believe this is essential as regulation of CSF and brain development is highly divergent across evolution, necessitating clarification and specificity of how genetics plays a role in human disease. Categories are defined a priori based on either phenotypic, molecular, or known genetic classifications. While many forms of HC can reasonably be classified into multiple categories, we attempt to simplify the groupings below.

Fig. 1
figure 1

Factors contributing to the development of hydrocephalus in humans

Fig. 2
figure 2

PRISMA flowchart outlining literature search to identify genes, mutations, and genetic mechanisms contributing to hydrocephalus in humans

Hydrocephalus secondary to aqueductal stenosis (AS)

Human genetics studies of HC secondary to aqueductal stenosis (AS) are summarized in Table 1. Fifteen unique gene mutations on 11 chromosomes inherited in both X-linked and autosomal patterns underlying HC secondary to AS have been identified. These genes include protocadherin 9 (PCDH9), immunoglobulin superfamily containing leucine rich repeat 2 (ISLR2), ATPase Na + /K + transporting subunit alpha 3 (ATP1A3), L1 cell adhesion molecule (L1CAM), FA complementation group C (FAC), fibroblast growth factor receptor 3 (FGFR3), solute carrier family 12 member 6 (SLC12A6), crumbs cell polarity complex component 2 (CRB2), Bardet-Biedl syndrome 7 (BBS7), podocin gene (NPHS2), multiple PDZ domain crumbs cell polarity complex component (MPDZ), laminin subunit beta 1 (LAMB1), alpha glucosidase (GAA), A-Disintegrin and Metalloproteinase with Thrombospondin motifs like 2 (ADAMTSL2), collagen type IV alpha 2 chain (COL4A2). A duplication in the Xp22.33 region and deletions of the long arm of chromosome 9, 12q22-q23.1, mutation in SRY-box transcription factor 2 (SOX2) gene, and mutation in the solute carrier family 12-member 7 (SLC12A7) gene were also identified.

Table 1 Aqueductal stenosis

Understanding the function of these genes may confer a mechanistic and phenotypic understanding of HC secondary to AS. For example, some patients with AS will display abnormal brainstem development leading to near complete obliteration of the aqueduct, whereas other children may display relatively normal anatomy associated with a web obscuring CSF flow. Genetics factors contributing to AS include ATP1A3, which encodes an ATPase ion channel that has been associated with CNS development and ventricular dilatation when disrupted in zebrafish [34]. In addition, SLC12A6 codes for the ion transporter KCC3 (K-Cl co transporter) that has been associated with AS among other phenotypes including peripheral neuropathy and agenesis of the corpus callosum in mice [35]. These ion channels are localized to the choroid plexus and are involved in neural stem cell development [36]. ADAMTSL2, encoding a glycoprotein, has been shown to interact with fibrillin 1 to enhance transforming growth factor- β (TGFβ) and fibroblast function. Additionally, TGFβ has been implicated in skeletal dysplasia and developmental dysfunction [37]. Thus, it is evident that genes with varying functions may contribute to AS and the diverse co-occurring phenotypes observed in these patients.

X-linked hydrocephalus

Genes contributing to X-linked HC include apoptosis inducing factor mitochondria associated 1 (AIFM1), adaptor related protein complex 1 subunit sigma 2 (AP1S2), EBP cholestenol delta-isomerase (EBP), FA complementation group B (FANCB), histone deacetylase 6 (HDAC6), OFD1 centriole and centriolar satellite protein (OFD1), OTU deubiquitinase 5 (OTUD5), coiled-coil domain containing 22 (CCDC22), and porcupine O-acyltransferase (PORCN). Table 2 summarizes the genetic studies of X-linked HC in humans. AIFM1 is involved in regulation of apoptosis [38]. In addition, AP1S2 regulates endosomal protein trafficking and structural integrity [39]. HDAC6 has been shown to interact with Runx2, a transcription factor involved in osteoblast differentiation, and other HDACs exhibit high expression patterns in prehypertrophic chondrocytes, indicating their role in endochondral ossification and skeletal dysplasias [40]. OTUD5 mutations also impact transcriptional regulation with its inability to prevent HDAC degradation and maintain neural stem cell development [41]. OFD1 and PORCN mutations affect signaling pathways such as hedgehog signaling or wingless/integrated (Wnt) signaling [42, 43].

Table 2 X-linked hydrocephalus

L1CAM associated hydrocephalus

Next, we discuss L1CAM associated HC, as this entity is well described and distinct phenotypically. Early linkage analysis studies of HC identified a mutation within the long arm of chromosome X, specifically Xq28. Further genomic analyses localized to a region between the gene loci of DXS52 and F8C, within which L1 cell adhesion molecule (L1CAM) resides. The genetic understanding of X-linked HC has primarily been linked to genetic alterations at the L1CAM locus. L1CAM duplications include the 3’ end of the open reading frame and exons 2–10. L1CAM insertions include exon 18 and the junction sequence between L1CAM and AVPR2. L1CAM deletions/microdeletions include exons 2, 5–8, 10, 11, 18, 19, 21–23, 26, intron 18, and whole gene deletion. L1CAM missense mutations include exons 1–16, 18, 20, 21, 24, 27, 28 and introns 2–4, 6–8, 10–15, 18, 21, 22, 24, and 26. L1CAM nonsense mutations include exons 1, 3, 8, 10–14, and 20–22. A silent mutation in Exon 8 of L1CAM has been associated with HC. A summary of all mutations across L1CAM can be found in Fig. 3 and Table 3. Mutations in L1CAM are also associated with MASA syndrome (characterized by mental retardation, aphasia, shuffling gait, and adducted thumb), and spastic paraplegia, highlighting the pleiotropic role of L1CAM in human disease.

Fig. 3
figure 3

L1CAM mutations implicated in human patients with hydrocephalus

Table 3 L1CAM

Dandy walker malformation

Dandy Walker malformation is a cerebellar structural anomaly that can impede CSF flow but can also be related to primary brain developmental alterations and contribute to HC development. Missense mutations are found in forkhead box C1 (FOXC1), fukutin (FKTN), laminin subunit gamma 1 (LAMC1), sphingosine-1-phosphate phosphatase 2 (SGPP2), and exocyst complex component 3 like 2 (EXOC3L2). Nonsense mutations are found in FKTN, nidogen 1 (NID1), and potassium channel tetramerization domain containing 3 (KCTD3). SIL1 nucleotide exchange factor (SIL1) displayed a nonstop mutation and carnitine palmitoyltransferase 2 (CPT2) displayed a deletion-insertion variant. Additional mutations included Zic family member 2 (ZIC2) and Zic family member 5 (ZIC5). Deletions were found in lysine methyltransferase 2D (KMT2D), chromosome 2 (2q36.1), chromosome 3 (3q25.1), chromosome 6 (6p24.1, 6p25.3), chromosome 7 (7p21.3), chromosome 8 (8q21), chromosome 12 (12q24), chromosome 13 (13q32), and chromosome 16 (16q21). The deletion of 8p21 resulted in the downregulation of fibroblast growth factor 17 (FGF17). Duplications were found in chromosome 6 (6p25.3), chromosome 7 (7p21.3), and chromosome 12 (12q24). In addition, EXOC3L2 regulates vesicular trafficking at synapses and cell polarity; a mutation within this gene locus can impact normal brain development [44]. KCTD3 is also highly expressed in the brain and kidneys and regulates ion channels such as hyperpolarization activated cyclic nucleotide-gated channel 3 (HCN3) [45]. SIL1 is a glycoprotein that regulates protein trafficking into the ER and ATPase activity, suggesting a mutated implication in protein folding through development [46, 47]. A patient with a mutation in CPT2, an enzyme responsible for breaking down long chain fatty acids, suggests a role of metabolic enzymes in the genetic susceptibility of HC secondary to Dandy Walker malformation [48]. Thus, Dandy Walker malformation related HC may be caused by a wide variety of genes involved in many biological processes. These data are summarized in Table 4.

Table 4 Dandy walker malformation

Ciliopathy

Genes involved in cilia function that are associated with HC are summarized in Table 5. Primary cilia dysfunction has been demonstrated to play a role in HC with numerous Mendelian ‘ciliopathies’ resulting in HC. Missense mutations were observed in Meckel-Gruber syndrome gene (MKS3), MKS transition zone complex subunit 1 (MKS1), intraflagellar transport 43 (IFT43), WD repeat domain 35 (IFT121), coiled-coil and C2 domain containing 2A (CC2D2A), transmembrane protein 216 (TMEM216), PKHD1 ciliary IPT domain containing fibrocystin/polyductin (PKHD1), intestinal cell kinase (ICK), exon 14 of KIAA0586, exons 4 and 13 of centrosomal protein 83 (CEP83), exons 6, 11, 12, 20, 23, 24, 28, 29, 32, and 36 of SET binding factor 2 (SBF2), exon 9 of zinc finger E-box binding homeobox 1 (ZEB1), and exon 5 of G protein subunit alpha i2 (GNAI2). Nonsense mutations were identified in CC2D2A, IFT121, forkhead box J1 (FOXJ1), exon 2 of KIAA0586, exon 3 of centrosomal protein 55 (CEP55), exons 3, 4, 7, and 13 of CEP83, and exon 11 of SBF2. Deletions and duplications resulting in frameshift mutations were found in CC2D2A, MKS3, MKS1, dynein axonemal intermediate chain 2 (DNAI2), IFT121, FOXJ1, exon 5 and 17 of CEP83, and exon 4 of ZEB1. Exon 2 was deleted in WD repeat-containing protein 16 (WDR16). Additional mutations were found in WD repeat domain 93 (WDR93). Loss of MKS3 and MKS1 are associated with ciliary shortening and dysfunction, suggesting a role in primary ciliary development. TMEM216 also contributes to ciliary development through apical polarization and formation and may result in Joubert, Meckel and related syndromes [49]. IFT43 and IFT121 maintain cilium organization and regulate intraflagellar transport in interaction with the IFT-A complex [50]. In addition, CEP83 also interacts with IFT proteins and guides vesicular docking ciliogenesis [51]. One patient was identified with a mutation in DNAI2, a component of the outer dynein arm complex (ODA), which is involved in cilia motility [52]. ZEB1, SBF2, and GNAI2 are involved in other signaling pathways previously identified in association with HC [53].

Table 5 Ciliopathy

PI3K-Akt-mTOR

Genes involved in PI3K-Akt-mTOR cell signaling pathway underlying HC are summarized in Table 6. Missense mutations were identified in ring finger protein 125 (RNF125), HECT and RLD domain containing E3 ubiquitin protein ligase family member 1 (HERC1), AKT serine/threonine kinase 3 (AKT3), mechanistic target of rapamycin kinase (mTOR), phosphatase and tensin homolog (PTEN), cyclin D2 (CCND2), phosphatidylinositol-4,5-bisphosphate 3-kinase catalytic subunit alpha (PIK3CA) (exon 18 and others), phosphoinositide-3-kinase regulatory subunit 2 (PIK3R2) (exon 13 and others), and platelet derived growth factor receptor beta (PDGFRB) (exon 12 and others). Deletions were observed in PIK3CA, and nonsense mutations were seen in PTEN [54]. Deletions in chromosome 1 (1q42.3-q44) resulted in the deletion of AKT serine/threonine kinase 3 (AKT3). Additional genetic mutations implicated in this pathway included those in tripartite motif containing 71 (TRIM71), SWI/SNF related matrix associated, actin dependent regulator of chromatin (SMARCC1), forkhead box J1 (FOXJ1), formin 2 (FMN2), patched 1 (PTCH1), and FXYD domain containing ion transport regulator 2 (FXYD2). Multiple genes within the PI3K-AKT-MTOR pathway highlight convergence on molecular mechanisms conferring risk to HC. Murine models have demonstrated the role of HERC1, which codes for an E3 ubiquitin ligase, to affect Purkinje cell physiology and mTOR activity [55]. TRIM71 and SMARCC1 are expressed within the ventricles and epithelium of mice brains (determined via in situ hybridization) suggesting that a mutation within this gene locus can affect this region may lead to HC [8]. Mutations in FOXJ1 and FMN2 have been shown to alter neuroepithelial integrity and lead to HC in mice [56, 57]. Mice harboring mutations in PTCH1 also display defects in ependymal cell integrity [58]. Thus, mutations within many genes converging on PI3K-Akt-mTOR signaling have been widely implicated in HC pathophysiology.

Table 6 PI3K-Akt-MTOR

Vesicle regulation & cell adhesion

Table 7 details mutations in genes responsible for vesicle regulation and cell adhesion that contribute to the development of HC. Missense mutations were found in and glial fibrillary acidic protein (GFAP). Sorting nexin 10 (SNX10) displayed a nonsense mutation and clathrin heavy chain (CLTC) displayed a frameshift mutation. Additional mutations include ArfGAP with FG repeats 1 (RAB), multiple PDZ domain crumbs cell polarity complex component (MPDZ), beta 1,3-glucosyltransferase (B3GALTL), SEC24 homolog D, COPII coat complex component (SEC24D), and actin beta (ACTB).

Table 7 Vesicle regulation and cell adhesion

GFAP is required for white-matter architectural development and myelination, perhaps accounting for the neurodevelopmental comorbidities frequently observed in patients with HC [59]. Mutations in the phosphoinositide binding domain of SNX10 alters endosomal integrity, suggesting a potential pathogenic mechanism in vesicular trafficking [60]. Additionally, mutations in this gene locus can disrupt interactions between sorting nexins and the V-ATPase complex further contributing to vesicle dysfunction and ciliopathy [61]. CLTC contributes to the development of the vesicular coat, and a mutation within this gene locus may disrupt vesicle stability [62]. Mutations in RAB27A are associated with Griscelli syndrome, characterized by albinism, hematological abnormalities, and organ malformation which can also present with HC [63]. SEC24 is also involved in intracellular trafficking by interacting with export signals from the endoplasmic reticulum and regulating cargo transport [64]. In addition, MPDZ is highly expressed in tight junctions suggesting that a mutation within this gene locus may disrupt alter tissue permeability [65]. Finally, B3GALTL interacts with the thrombospondin type 1 repeat (TSR) protein family which play varied roles in maintain and regulating cell–cell adhesion [66].

Glycosylation defects

Table 8 summarizes genes implicated in human HC associated with defects in glycosylation. Nonsense mutations were seen in protein O-mannose kinase (POMK), and protein O-mannosyltransferase 1 (POMT1). Loss of function mutations were identified in dystroglycan 1 (DAG1) and isoprenoid synthase domain containing gene (ISPD). Additional mutations included those in protein C, inactivator of coagulation factors Va and VIIIa (PROC), fukutin related protein (FKRP), protein O-mannosyltransferase 2 (POMT2), protein O-linked mannose N-acetylglucosaminyltransferase 1 (beta 1,2-) (POMGNT1), LARGE xylosyl and glucuronyltransferase 1 (LARGE1) and a translocation between chromosome 5 and 6, t(5;6) (q35;q21). In addition, DAG1 codes for dystroglycan, a protein involved in extracellular matrix integrity and the genetic etiology of many neurological syndromes. Mutations in DAG1 have been found to contribute to Walker-Warburg syndrome and other muscular dystrophy-dystroglycanopathies which can be associated with HC [67]. Dystroglycan may also be affected through defects in its glycosylation patterns. For instance, mutations in POMK have been shown to impair the glycosylation of a-dystroglycan affecting cytoskeleton stability [68]. Other genes contributing to dystroglycanopathies through glycosylation errors include POMT1, POMT2, POMGNT1, FKRP, ISPD and LARGE1 [69].

Table 8 Glycosylation defects

Growth factor related signaling

Table 9 summarizes genetic mutations associated with growth factor related signaling dysfunction. Mutations were observed in fibroblast growth factor receptor 1 (FGFR1), fibroblast growth factor receptor 2 (FGFR2), fibroblast growth factor receptor 3 (FGFR3), ZPR1 zinc finger (ZPR1), and fibrillin 1 (FBN1). Specifically, exon 7 displayed a missense mutation in FGFR2 and exon 64 displayed a mutation in FBN1. Mutations in FGFR play pleiotropic roles in numerous syndromes including Crouzon syndrome, Jackson-Weiss syndrome, Apert syndrome and Pfeiffer syndrome [70,71,72,73]. These craniosynostoses have been associated with HC and FGFR mutations contributing to bony abnormalities, which may explain the venous and CSF outflow obstructions leading to this phenotype [74]. The FGFR mutations identified are predominantly gain of function mutations altering ligand binding and tyrosine kinase activity [75]. In addition, ZPR1 contributes to cell proliferation and FBN1 is associated with TGF beta signaling suggesting their mechanistic contributions to the HC phenotype seen in patients with these phenotypes [76].

Table 9 Growth factor signaling

Extracellular matrix defects

Table 10 highlights the genetic mutations contributing to extracellular matrix defects. Mutations were found in fukutin (FKTN), cartilage associated protein (CRTAP), collagen type VIII alpha 2 chain (COL8A2), collagen type III alpha 1 chain (COL3A1), collagen type IV alpha 1 chain (COL4A1), vascular cell adhesion molecule 1 (VCAM1), protein tyrosine phosphatase receptor type F (PTPRF), fibrillin 1 (FBN1), laminin subunit beta 1 (LAMB1), FRAS1 related extracellular matrix 1 (FREM1), and the plasminogen gene. CRTAP is involved in proline hydroxylation which ultimately contributes to collagen stability and functionality [77]. Mutations within the CRTAP gene locus can lead to Cole-Carpenter syndrome, which is associated with HC [77]. Other basement membrane proteins encoded by COL8A2, COL3A1, COL4A1, VCAM1, and PTPRF may alert the extracellular matrix and contribute to HC. For instance, a mutation in COL3A1 affects its triple helix stability leading to degradation and further defects in the basement membrane [78]. LAMB1 knockdown in zebrafish disrupted laminin integrity, a component of the basal lamina, leading to brain structural abnormalities [79], suggesting a potential pathogenic link to HC.

Table 10 Extracellular matrix defects

Neurogenesis and neural stem cell biology

Table 11 summaries gene mutations implicating neurogenesis. Mutations were identified in SRY-box transcription factor 9 (SOX9), solute carrier family 29 member 3 (SLC29A3), adhesion G protein-coupled receptor (ADGRG1), katanin interacting protein (KIAA0556), G protein signaling modulator 2 (GPSM2), tripartite motif containing 71 (TRIM71) [80], SWI/SNF related, matrix associated, actin dependent regulator of chromatin subfamily c member 1 (SMARCC1) [81], patched 1 (PTCH1), FLVCR heme transporter 2 (FLVCR2), intestinal cell kinase (ICK), cystathionine beta-synthase (CBS), 5-methyltetrahydrofolate-homocysteine methyltransferase reductase (MTRR), interleukin 4 induced 1 (IL4I1), scribble planar cell polarity protein (SCRIB1), protein tyrosine kinase 7 (PTK7), frizzled class receptor 1 (FZD1), VANGL planar cell polarity protein 2 (VANGL2), dishevelled segment polarity protein (DVL2), transcription elongation factor B polypeptide 3B (TCEB3B), phospholipase C delta 4 (PLCD4), Ras associated domain family member 4 (RASSF4), phenylalanyl-tRNA synthetase 2, mitochondrial (FARS2), tubulin beta 3 class III (TUBB3), and discs large MAGUK scaffold protein 5 (DLG5). Frameshift mutations were seen in WD repeat domain 81 (WDR81), kinase D interacting substrate 220 (KIDINS220). Deletions were seen in chromosome 6 (6q25.3 and 6p25), chromosome 13 (13q), chromosome 16 (16p12.2), and chromosome 22 (22q11.2). The 6p25 deletion resulted in the deletion of forkhead box C1 (FOXC1), forkhead box F2 (FOXF2), and forkhead box Q1 (FOXQ1).

Table 11 Neurogenesis

The heterogeneity of neurogenesis-associated HC suggests that numerous genes involved in development may confer susceptibility to this phenotype. SOX9 knockdown in mice suggest a role in neural stem cell development and ependymal cell maintenance as a pathogenic mechanism that causes HC [82]. In addition, mutations in ADGRG1 have been shown to impact cerebral cortex development and neuronal migration via the perturbation of the RhoA pathway [83]. In addition, GPSM2 has been shown to alter neuroepithelial function through disruption of cellular orientation and planarity leading to aberrant brain development [84]. Finally, mice lacking KIDINS220 display attenuated responses to neurotrophic factors and have impaired development in multiple signaling pathways [85]. Understanding the genetic influence of neurogenesis may elucidate a better understanding of patient characteristics and poor outcomes in the HC phenotype [86,87,88,89].

Inherited cancer syndromes

Table 12 summarizes genes that contributed to tumor pathogenesis, and which result in the development of HC. Mutations are seen in NRAS proto-oncogene, GTPase (NRAS), von Hippel-Lindau tumor suppressor (VHL), patched 1 (PTCH1) and FA complementation group C (FANCC). Germline mutations are seen in phosphatase and tensin homolog (PTEN) and SUFU negative regulator of hedgehog signaling (SUFU). Deletions within chromosome 11 (11p13) and chromosome 9 (9q22.3 and 9q22-q31) were also identified. NRAS is an oncogene contributing to the development of congenital melanocytic nevi, a condition associated with HC [90]. Clinically relevant mutations in Von Hippau Lindau (VHL) affect protein expression and degradation where patients with or without a mass lesion (i.e., hemangioblastoma) develop HC [91]. Gorlin syndrome is disorder characterized with bony abnormalities and an increased risk for multiple CNS and non-CNS tumors. Previous studies have mapped this syndrome to deletions in the 9q22 locus which is consistent with the patients identified in this review with mutations specifically affecting PTCH1 and FANCC genes [92, 93]. Finally, mutations in SUFU have also been associated with Gorlin syndrome [94, 95].

Table 12 Inherited cancer syndromes

WNT signaling

WNT signal transduction is involved in numerous pathways regulating cell function and development. Table 13 summarizes gene mutations identified in HC patients with this pathway. Numerous studies have reported gene mutations in coiled-coil and C2 domain containing 2A (CC2D2A) and coiled-coil domain containing 88C (CCDC88C). COACH syndrome is defined as cerebellar vermis hypoplasia, oligophrenia, ataxia, colobomas, and hepatic fibrosis [96]. This gene locus has been shown to interact with the WNT signaling pathway and is associated with centrosome stability [97]. In addition, CCDC88C is associated with the WNT signaling pathway through interaction with the Dishevelled protein [98]. The dishevelled protein contains a binding domain which interacts with a hook related protein transcribed from the CCDC88C locus [99]. WNT signaling plays numerous roles in cell communication and embryonic development, suggesting potential mechanisms contributing to HC [100].

Table 13 WNT signaling

Transcriptional, post-transcriptional, and epigenetic regulation

Table 14 summarizes mutations in genes that regulate transcription, post-transcriptional, and epigenetic processes. Missense mutations were seen in THO complex subunit 6 (THOC6) and HYLS1, genes involved in transcriptional regulation. Patients with loss of function mutations in FA complementation group L (FANCL) we identified. Additional mutations observed included interferon regulatory factory 6 (IRF6), small nucleolar RNA, C/D box 118 (SNORD118), nuclear factor I A (NFIA), SET binding protein 1 (SETBP1), SWI/SNF related, matrix associated, actin dependent regulator of chromatin, subfamily b, member 1 (SMARCB1), maelstrom spermatogenic transposon (MAEL), a deletion in chromosome 5 (5q35.3), and 20q13.3 trisomy. Deletions in chromosome 1 (1q42.3-q44) resulted in the deletion of zinc finger and BTB domain containing 18 (ZBTB18) and heterogeneous nuclear ribonucleoprotein U (HNRNPU). THOC6 is a part the TREX complex responsible for mRNA export and is localized to the 5’ cap of mRNA [101]. It has been associated with Beaulieu-Boycott-Innes syndrome, which is charactered by developmental delay and organ dysgenesis [102]. Patients identified in this review with Beaulieu-Boycott-Innes syndrome and THOC6 mutations have been shown to develop HC, suggesting a role for mRNA export regulation in association with HC phenotypes [103]. HYLS1 is associated with Hydrolethalus syndrome, a disorder characterized by HC and craniofacial abnormalities [104]. Expression analysis of this gene suggests a role in CNS development, where a HC associated mutation gene causes nuclear localization whereas the WT form is expressed in the cytoplasm [104]. SNORD118 is involved in regulation of ribosome biology and associated with the hydrocephalic phenotype of Labrune syndrome, characterized by leukoencephalopathy, intracranial cysts, and calcification [105]. While the function of SETBP1 remains largely unknown, mutations in this gene are associated with Schinzel-Giedion syndrome, characterized by facial abnormalities, intellectual disability, congenital malformations, and HC [106]. SMARCB1 is involved in chromatin remodeling to further enhance or repress transcription [107]. Finally, a transcriptome-wide association study (TWAS) and multi-omics study of HC identified maelstrom (MAEL), a gene that regulates transposons and epigenetic modifications, as an experiment-wide predictor of HC in the cortex [9, 108]. These studies identified transcriptional regulators and further emphasize the need to explore these mechanisms to understand the mechanistic associations with HC.

Table 14 Transcriptional, post-transcriptional, and epigenetic regulation

Ion transport and regulation

Table 15 summarizes gene mutations implicating ion transport. Mutations were seen in aquaporin 4 (AQP4) and FLVCR heme transporter 2 (FLVCR2). Mutations on chromosome 17 (17p13) implicated transient receptor potential cation channel subfamily V member 3 (TRPV3). Aquaporin 4 (AQP4) regulates water transport on ependymal cells and knockout of this gene in mice show disrupted gap junctions which alter the ependymal zone and CSF flow contributing to HC development [109; 110]. Mutations in the enhancer of TMEM50b alter expression of TTF, a direct transcriptional regulator of AQP1, have also been identified [101]. Mutations in FLVCR2 are associated with Fowler’s syndrome, a disorder characterized by HC and hydranencephaly [111]. This gene locus encodes a transmembrane protein involved in solute transport, suggesting that defects in chemiosmotic regulation contribute to HC development [112].

Table 15 Ion transport and regulation

Normal pressure hydrocephalus

Normal pressure HC (NPH) is a form of communicating HC in which the progressive pressure of CSF is believed to result in in ventricular dilatation and further CSF accumulation. Table 16 summarizes the genes implicated in human studies of NPH. Scm like with four mbt domains 1 (SFMBT1) displayed an intron 2 deletion. Cilia and flagella associated protein 43 (CFAP43) was found to have a nonsense mutation. The gene locus contributing to the development of ETINPH, a disorder characterized with essential tremors and idiopathic NPH, was localized to 19q12-13.31 on chromosome 19. SFMBT1 is highly expressed in ependymal cells and epithelial cells of the brain, suggesting that a mutation in this gene locus may contribute to the dysfunctional CSF circulation [113]. Furthermore, a binding site had been identified within intron 2 of this gene locus, suggesting that the deletion of this intron, as seen in our review, will impact function [114; 115]. Deletion of cell wall biogenesis 43 (CWH43) in humans has also been associated with NPH [116].

Table 16 Normal Pressure Hydrocephalus

Metabolism

Table 17 indicates genes involved in metabolic pathways. Mutations were seen in cytochrome c oxidase subunit 6B1 (COX6B1), methylenetetrahydrofolate reductase (MTHFR), and sulfatase modifying factor 1 (SUMF1). Mutations in COX6B1 have been shown to disrupt the electron transport chain suggesting that alterations in cellular energetics can contribute to HC [117] [118]. SUMF1 encodes formylglycine generating enzyme (FGE) involved in modifying cysteine residues in the endoplasmic reticulum [119]. MTHFR regulates folate metabolism, and mutations within this gene locus have been identified in congenital HC patients providing rationale to explore metabolic genes and their association with pathology [120].

Table 17 Metabolism

Cell cycle and cytoarchitecture

Table 18 displays genes involved in cell cycle regulation and cytoarchitecture. Mutations were seen in spindle apparatus coiled-coil protein 1 (SPDL1), tubulin alpha 3e (TUBA3E), nidogen 1 (NID1), tRNA splicing endonuclease subunit 15 (TSEN15), clathrin heavy chain linker domain containing 1 (CLHC1), TBC1 domain containing kinase (TBCK), xin actin binding repeat containing 1 (XIRP1), nucleoporin 107 (NUP107), erythrocyte membrane protein band 4.1 like 4A (EPB41L4A), protein phosphatase 2 regulatory subunit B delta (PPP2R5D), protein phosphatase 2 scaffold subunit Alpha (PPP2R1A), prolyl 4-hydroxylase subunit beta (P4HB), and crumbs cell polarity complex component 2 (CRB2). SPDL1 has been shown to regulate mitotic checkpoints, and mutations arrested affected cells in metaphase [121]. TUBA3E maintains microtubule integrity by encoding for part of the microtubule heterodimer, alpha tubulin [122]. TSEN15 contributes to an endonuclease complex involved in tRNA splicing, and mutations affecting this gene locus can lead to defects in cell division [123]. XIRP1 has been shown to maintain actin integrity and stability [124]. P4HB encodes an enzyme subunit involved in collagen formation, and mutations affecting this gene location are associated with reduced cytoarchitectural stability [125].

Table 18 Cell cycle and cytoarchitecture

Lipid structure and regulation

Table 19 summarizes genes involved in lipid structure and regulation associated with HC in humans. Mutations were seen in bridge-like lipid transfer protein family member 1 (KIAA1109), and glucosylceramidase beta 1 (GBA). The KIAA1109 ortholog in Drosophila melanogaster has shown to affect synaptic growth at the neuromuscular junction through modulation of phosphatidylinositol 4,5-bisphosphate (PIP2) [126]. GBA encodes for a lysosomal enzyme responsible for metabolizing glycolipids [127].

Table 19 Lipid structure and regulation

Genes of unknown function

Table 20 summarizes genes that are associated with HC pathology without a clear function. Additional variants include partial 1q trisomy, tetrasomy 5p, tetraploidy of chromosome 9, trisomy 9p, and chromosome 21 trisomy. Studies that have identified mutations in chromone 6 displayed microdeletions or mosaicism of monosomy. Deletions in chromosome 8 (8q12.2-q21.2) and chromosome 16 (16q) were also identified, and microduplications in chromosome 17 (17p13.1) have been reported. The vast genetic influence on HC emphasizes importance of exploring and understanding the factors that confer genetic risk to improve diagnostic and prognostic efficiency. Autosomal and sex chromosomal location of all genetic findings included in this review is summarized in Figs. 3, 4.

Table 20 Genes of unknown function
Fig. 4
figure 4

Chromosome map of hydrocephalus-associated loci across autosomal and sex chromosomes in humans

Global burden of genetic hydrocephalus

We next aimed to quantify the country of origin for patients included in this review (Fig. 5). Given the wide range of HC disease burden across the world [3], we aimed to determine if genetic studies of HC were similarly representative. What is clear is that for regions of the world where HC prevalence is highest (Africa, East Asia, etc.), there is an obvious lack of genetic studies of HC of any kind. For example, there is not a single study performed by authors in Sub-Saharan Africa (SSA) or including people from SSA. Given that African genomes are the most diverse and complex with generations of environmental pressures (including emerging pathogens) shaping the genome, understanding genetic risk factors in these populations is essential. While epidemiological estimates of the contribution of genetically linked forms of HC is not feasible at present, these data begin to highlight disparities in representation of genetic studies and the need for large-scale genetic studies of HC in diverse populations. These data also provide a reasonable estimate of the potential burden, although likely underestimated, contribution of genetic factors contributing to HC. It is our hope that this review highlights the diverse mechanisms underlying HC, the complex molecular pathways that may contribute to HC pathogenesis, and the need to greatly expand the representation of diverse peoples in HC genetics research.

Fig. 5
figure 5

Heatmap of the globe demonstrating the country of origin for patients with genetic contributions to hydrocephalus. Figure created with OpenStreetMap

Discussion

HC is a complex, heterogenous condition that can be a component of a wide range of genetic conditions and can be caused by a variety of preceding environmental factors. Because HC is a component of many syndromes with a wide range of concomitant phenotypes, understanding the genetic pleiotropism of contributing genes is important for delineating the pathophysiologic basis of the disease. This review provides a broad overview of the associations between genetic mechanisms underlying HC. The variability in phenotypes observed, methodology used to uncover genetic information, and wide range of validation of genetic findings highlights the major challenges in the field. While many studies are descriptive, a wide range of hypotheses are generated based on implicated genes and potential mechanisms. Specifically, many studies implicate alterations in neurogenesis and primary brain development, as opposed to direct alterations in CSF regulation, as potential pathophysiologic mechanisms. Overall, as genomic technologies become more ubiquitous in clinical practice and more patients undergo unbiased genomic sequencing, our understanding of HC will improve. However, there are several limitations and points to consider as this field evolves.

An ongoing challenge in human genetics is proving causality of implicated genetic findings. Classical validation technique requires reproducing the implicated mutation (if evolutionarily conserved) in a model organism such as a mouse or rat. However, the physiologic regulation of CSF and mechanisms underlying brain development are markedly different in these model organisms and often do not recapitulate human disease. Many genes underlying HC are associated with other phenotypes, and it may not be possible to identify a secondary causative genetic factor that unmasks the phenotype. Since this approach relies on the gene product being evolutionarily conserved, identification of human-specific disease mechanisms is impaired. Alternative approaches to determining the relationship of a gene variant to a HC phenotype include structural biologic modeling of presumed deleterious mutations; however, this approach does not consider physiological and phenotypic heterogeneity. Similarly, these approaches often rely on protein expression in prokaryotic systems, limiting interpretation of post-translational modifications and other physiologic contributors to protein function.

Based on the significant co-occurrence of traits affecting other organ systems, it is likely that genes associated with HC display significant pleiotropy. A simplistic model of monogenic contributions to HC is unlikely to capture the genetic etiology of most cases. Even among monogenic contributions to HC, there is significant phenotypic and genetic variability (i.e., L1CAM). As quantitative genetic methodology improves to identify polygenic contributors to disease, we suspect that a much larger proportion of cases will have polygenic contributions. Because HC is a heterogenous disease, accrual of large numbers of ‘homogenous’ cases are needed to accurately quantify reproductible genetic associations.

The variability in genomic technology used to determine potential genetic contributions to HC is significant. Agnostic methods such as genome wide association studies (GWAS), transcriptome-wide association studies (TWAS), whole-exome sequencing (WES), and whole-genome sequencing (WGS) have been used, but are limited by cost, sample size, and technical expertise involved in analysis. In contrast, targeted sequencing approaches rely on hypothesis-driven identification of implicated genetic loci introducing significant experimental bias.

Our review highlights that most genetic studies of HC are performed in countries where disease burden, paradoxically, is amongst the lowest in the world. This reflects disproportionately low resources for genetic studies in low- and middle-income countries. For example, Sub Saharan Africa the most genetically diverse and complex region in the world, where the burden of HC is also the highest, yet there are no genetic studies of HC of any kind in these populations. Although the burden of HC is largely the result of infections, the genetic contributors to infection susceptibility are largely uncharacterized in these populations. Evolutionary selection pressures have been differentially shaped by exposure to infectious pathogens, geographic shifts of ancestral peoples, and population isolation. Therefore, understanding genetic factors specific to these populations is paramount to improve secondary prevention and moving towards non-surgical treatment options.

Advances in genetic technology and interpretation coupled with decreased costs will garner a new era of precision medicine that can be applied in the clinic [128]. The extent to which genetic information may guide treatment in HC has not been fully realized. As more patients are rigorously studied using complementary and convergent genomic approaches coupled with long-term clinical outcomes, we may be able to incorporate genetic information into clinical care. Owing to the genetic architecture of HC highlighted here and across many studies, we anticipate that creation of polygenic risk scores (PRS) may be the most clinically meaningful and practical for disease prognostication and understanding comorbid disease risks.

Conclusions

HC is a phenotypically and genetically complex disease. While the literature describing the genetic causes of HC is vast, this comprehensive review highlights opportunities for further mechanistic study and disparities in ancestral representation. The varying rigor with which genetic studies are conducted highlights the challenge of determining causality of implicated genomic alterations, inadequacies of current model systems, and the need for human-specific molecular validation studies. What is clear is that our genetic understanding of HC is incomplete and our understanding of pleiotropy of implicated HC genes requires further maturation. This study represents the first large-scale systematic literature review of the genetic basis of HC in humans and highlights many areas ripe for future investigation.