The prevalence and phenotypic range associated with biallelic PKDCC variants

Abstract PKDCC encodes a component of Hedgehog signalling required for normal chondrogenesis and skeletal development. Although biallelic PKDCC variants have been implicated in rhizomelic shortening of limbs with variable dysmorphic features, this association was based on just two patients. In this study, data from the 100 000 Genomes Project was used in conjunction with exome sequencing and panel‐testing results accessed via international collaboration to assemble a cohort of eight individuals from seven independent families with biallelic PKDCC variants. The allelic series included six frameshifts, a previously described splice‐donor site variant and a likely pathogenic missense variant observed in two families that was supported by in silico structural modelling. Database queries suggested that the prevalence of this condition is between 1 of 127 and 1 of 721 in clinical cohorts with skeletal dysplasia of unknown aetiology. Clinical assessments, combined with data from previously published cases, indicate a predominantly upper limb involvement. Micrognathia, hypertelorism and hearing loss appear to be commonly co‐occurring features. In conclusion, this study strengthens the link between biallelic inactivation of PKDCC and rhizomelic limb‐shortening and will enable clinical testing laboratories to better interpret variants in this gene.


Funding information
Medical Research Council, Grant/Award Number: MR/W01761X/1; NIHR Oxford Biomedical Research Centre; Wellcome Trust, Grant/Award Number: 203141/Z/16/Z been implicated in rhizomelic shortening of limbs with variable dysmorphic features, this association was based on just two patients. In this study, data from the 100 000 Genomes Project was used in conjunction with exome sequencing and panel-testing results accessed via international collaboration to assemble a cohort of eight individuals from seven independent families with biallelic PKDCC variants. The allelic series included six frameshifts, a previously described splice-donor site variant and a likely pathogenic missense variant observed in two families that was supported by in silico structural modelling. Database queries suggested that the prevalence of this condition is between 1 of 127 and 1 of 721 in clinical cohorts with skeletal dysplasia of unknown aetiology. Clinical assessments, combined with data from previously published cases, indicate a predominantly upper limb involvement. Micrognathia, hypertelorism and hearing loss appear to be commonly co-occurring features. In conclusion, this study strengthens the link between biallelic inactivation of PKDCC and rhizomelic limb-shortening and will enable clinical testing laboratories to better interpret variants in this gene. Where novel disease-gene associations are proposed based on so few patients it is important for additional families to be described, even where there is support from pathway analysis and model organisms. Consequently, PKDCC was initially classed as 'Amber' in the skeletal dysplasia panel on PanelApp. Variants in PKDCC were therefore not prioritised by the 100 000 Genomes Project 6 (100 kGP) and other studies that utilised PanelApp as part of their pipeline.
In this study, genome, exome and panel-NGS sequencing was performed in tandem with international collaborative efforts to assemble a cohort of seven new families with rare biallelic variants in PKDCC. As four of the families were recruited via large clinical databases, this allowed us to estimate the prevalence of this rare condition in suitable patient cohorts. These results strengthen the reported gene-disease association and in combination with the cases published previously, enable a more precise delineation of the phenotypic spectrum associated with this rare condition.

| METHODS
Families 1 and 2 were sequenced as parent-child trios as part of the 100 kGP. Ethics approval was from Cambridge South REC (14/EE/1112) and the study conforms to recognised standards. Library preparation was using the TruSeq PCR-free high throughput kit.
Genome sequencing data was generated in a single lane of a HiSeqX (Illumina) and 150 bp paired-end reads were mapped using the iSAAC aligner to GRCh38. Severe consequence homozygous variants were extracted using the 'tiering_data' table available within Genomic England's research environment and a 1% population allele frequency threshold was employed. Families 3 and 4 were identified via Gene-Matcher (https://genematcher.org). Exome sequencing was performed using the IDT xGen Exome Research Panel v1.0 by GeneDx, as described. 7 In Family 5, customised exome sequencing was performed using the Illumina NovaSeq platform (Data S1). For Family 6, sequencing had been performed at up to 50x using NGS-panels comprising 438 genes. The individual from Family 7 underwent clinical exome analysis, as described. 8 Structural modelling is described in Data S1.

| RESULTS
Using the tiering data generated as part of the 100kGP, we Initially, no further biallelic PKDCC cases from the 100kGP could be identified. However, following discussion at a multidisciplinary team meeting to review unsolved musculoskeletal cases, PKDCC emerged as a strong candidate in Family 2. The proband (F2-II-1) presented isolated rhizomelia of upper limbs. Non-skeletal features included mild neutropenia, hypoplastic pituitary gland on imaging, growth hormone deficiency and low TSH. Teeth showed evidence of enamel pitting and there was mild hypertelorism and a slight metopic ridge. In utero radiological assessment at 20 weeks also indicated that the proband was likely to have a similarly affected sibling and rhizomelic limb-shortening has now been confirmed postnatally. This last observation made an autosomal-recessive mode of inheritance more likely. PKDCC was one of 70 genes present on the current PanelApp skeletal dysplasia gene list (https://panelapp.genomicsengland.co.uk/ panels/309), but absent from the version used at the time of the initial analysis (Table S1). Finally, PKDCC lay in a 791 kb ROH (Table S2,  septal defect, micrognathia, hypertelorism, cupped ears with rotated, fleshy ear lobes, an elfin-like face and rhizomelia. Inferior vermis hypoplasia was also noted but asymptomatic. Family history is significant for a similarly affected older brother. In all families, variant segregation was consistent with an autosomalrecessive mode of inheritance ( Figure 1). Of the eight variants observed across these seven families, all have been validated ( Figure S2) and have allele frequencies in gnomAD v2.1.1 of 0-6/158918 (Table S3) (Table S3).
Differential diagnoses for this condition may also include Robinow syndrome, which was proposed as a clinical diagnosis in F1/F6/F7. Proptosis, depressed nasal bridge, thick arched eyebrows, a long philtrum and thin upper lips were seen in the younger patients but absent in the adult patient, which could suggest the evolution of the facial features with age ( Figure 2F-K). Hearing loss was reported for both sisters in Family 3 and for F5-II-10. This feature was also seen in one of the cases reported previously 5 suggesting this to be a recurring characteristic of this condition. Downslanting palpebral fissures, brachial cleft defect and hoarse voice, were seen in F5-II-10 and was previously reported in one of the patients of Sajan et al. Sloping shoulders were seen in three of the patients reported here. Short thumbs and clinodactyly or shortening of the third/fifth digit were seen in multiple patients. In the lower limbs, patellofemoral joint dislocation was observed in three cases and genu varum was seen in F5-II-10. Notably, this individual is the first adult case to be described and a more detailed clinical summary is provided in Data S1. Following radiological assessment, short humeri were noted in the skeletal surveys available for 6 of 8 cases and radiological data was available for review in five families ( Figure 2L-T).
One limitation of this study is that for p.(Leu262Arg), further evidence supporting pathogenicity would be desirable. Although in silico analysis predicted that the variant may impact protein stability, this should ideally be confirmed experimentally. Segregation studies on F5-II-10's 9 siblings (should DNA become available) may also be used to help support pathogenicity. The same variant in Family 6 was originally assessed as a VUS (ClinVar, SCV002127124.1). However, based on clinical judgement and the finding of the same variant in F5, the patient is now being treated as if they have a molecular diagnosis.
As four of the families were recruited from large clinical databases, we were able to estimate the likely prevalence of this condition in suitable patient populations. There was a 2 of 253 incidence of PKDCC positive cases amongst patients recruited to the 100 kGP (v15, 26 May 2022) due to unexplained skeletal dysplasia. Families 3 and 4 were identified from the GeneDx variant database (v14 April 2021) containing exome data for 1442 individuals with skeletal dysplasia phenotypes. In combination (4/1695), and given that skeletal T A B L E 1 Summary of clinical findings and variants across the 10 individuals with PKDCC variants (8 described here and 2 by Sajan et al. 5   Three of the variants identified lie in a low complexity GC-rich region (p.Gly74-Gly131). In Family 2, this contributed to p.(Pro77Alafs*95) being called as a low-confidence variant ('SC; badReads'), explaining why it had initially been missed. Genomic regions with low complexity can be highly mutagenic and are often hard to analyse. We speculate that patients with upper-limb rhizomelia in whom a single heterozygous pathogenic allele in PKDCC has been uncovered may benefit from closer scrutiny of this region.
In summary, we identified eight affected individuals from seven independent families with biallelic loss of function variants in PKDCC.
These data help delineate the phenotypic spectrum associated with this ultra-rare condition and will allow clinical laboratories to have more confidence in interpreting variants in this gene.