Genetic landscape of congenital insensitivity to pain and hereditary sensory and autonomic neuropathies

Abstract Congenital insensitivity to pain (CIP) and hereditary sensory and autonomic neuropathies (HSAN) are clinically and genetically heterogeneous disorders exclusively or predominantly affecting the sensory and autonomic neurons. Due to the rarity of the diseases and findings based mainly on single case reports or small case series, knowledge about these disorders is limited. Here, we describe the molecular workup of a large international cohort of CIP/HSAN patients including patients from normally under-represented countries. We identify 80 previously unreported pathogenic or likely pathogenic variants in a total of 73 families in the >20 known CIP/HSAN-associated genes. The data expand the spectrum of disease-relevant alterations in CIP/HSAN, including novel variants in previously rarely recognized entities such as ATL3-, FLVCR1- and NGF-associated neuropathies and previously under-recognized mutation types such as larger deletions. In silico predictions, heterologous expression studies, segregation analyses and metabolic tests helped to overcome limitations of current variant classification schemes that often fail to categorize a variant as disease-related or benign. The study sheds light on the genetic causes and disease-relevant changes within individual genes in CIP/HSAN. This is becoming increasingly important with emerging clinical trials investigating subtype or gene-specific treatment strategies.

overcome limitations of current variant classification schemes that often fail to categorize a variant as disease-related or benign.The study sheds light on the genetic causes and disease-relevant changes within individual genes in CIP/HSAN.This is becoming increasingly important with emerging clinical trials investigating subtype or gene-specific treatment strategies.

Introduction
Complex genetic variability leads to individual differences in the perception of pain.In contrast to polygenic and environmental correlations, specific single nucleotide variants can have an effect such that the sensation of pain is absent from birth, or a progressive loss of pain sensitivity becomes apparent in the course of life.In these rare and monogenic diseases, there is usually a developmental disorder of pain-sensing neurons, neurodegeneration of peripheral nerves, or altered electrical activity of nociceptors.This heterogeneous group of genetic pain loss disorders includes congenital insensitivity to pain (CIP), hereditary sensory neuropathy (HSN) and, if autonomic nerves are involved, hereditary sensory and autonomic neuropathy (HSAN).The HSNs are assigned here to the group of HSAN diseases.The consequences of pain loss are recurrent injuries and fractures resulting in mutilation or amputation, often in combination with severely impaired wound healing.The sensation of itch, temperature and touch may also be impaired with negative impact on health.Affected patients can have marked autonomic dysfunction, such as anhidrosis, gastrointestinal and sexual dysfunction or blood pressure fluctuations.In some subtypes of CIP/HSAN, patients also show intellectual disability, muscle weakness, ataxia or other additional symptoms.To date, pathogenic variants in more than 20 genes are known to cause pain loss syndromes.Various cellular processes can be affected, including sodium channel activity, 1,2 sphingolipid metabolism, 3,4 membrane dynamics, [5][6][7] axonal transport, 8,9 neurotrophin signalling, 10,11 epigenetic regulation 12 or cytoskeletal architecture. 13Because of the rarity of CIP/HSAN disorders, knowledge of these conditions is limited and diagnosis is often delayed or incorrect resulting in a diagnostic odyssey.Moreover, except for few studies including larger patient numbers, 14,15 the literature is often restricted to single case descriptions.][18] In this retrospective study, we provide deeper insights into the clinical and genetic landscape of these rare conditions by sequencing of the as yet largest cohort of CIP/HSAN patients.

Patient cohort
Genetic data were collected retrospectively from existing datasets from patients who had been either referred directly to the participating centres or whose blood samples and clinical information were sent to the participating centres.All patients showed clinical signs of HSAN or CIP (i.e.reduced sensation of pain, temperature and touch either congenital or developed later in life and/or clinical manifestations such as unnoticed injuries, skin ulcerations, amputations, osteomyelitis, painless fractures).Since neuropathic pain, especially in the initial stage, may be a sign of different forms of HSAN, patients with a combined phenotype and a variant in one of the HSAN-related genes were also included in the study.Patients with suspected other genetic disorders potentially mimicking the phenotype of HSAN, such as Charcot-Marie-Tooth disease (CMT) or e.g.Lesch-Nyhan syndrome as a typical example with self-mutilating behaviour, were not included in the study.Other possible underlying non-genetic causes of decreased pain sensitivity (e.g.toxic, metabolic or infectious causes of polyneuropathy) were queried by the respective clinical centres and resulted in exclusion from the study.The study was conducted in accordance with the Declaration of Helsinki and has been approved by the local ethics committees of the participating institutions.Prior to inclusion, written informed consent was obtained from patients or their legal guardians

Short-read next-generation sequencing
Project sites and collaborating research institutions provided genetic data of CIP/HSAN patients that had been analysed by short-read sequencing.Datasets were analysed with regard to novel variants in known CIP/HSAN genes [genes from the panel 'pain syndromes' v1.12, Genomics England Panel App were prioritized (https:// panelapp.genomicsengland.co.uk/panels/288/)].Variant calling was done by each research institute separately; protocols, consumables and pipelines used differed between the institutions.Detailed protocols can be provided upon request.Cases were considered if mono-or biallelic variants (for dominant and recessive disorders, respectively) were found in one of the core genes with no additional probably disease-associated variants detected by screening of the datasets.The MasterMind database was checked in June 2023 and variants were included if they (i) had not been published in the literature at all; (ii) had only been described in supplementary materials; or (iii) had only been reported in patients with a phenotype other than CIP/HSAN (Supplementary Table 1).If possible, segregation analyses within the families were performed using Sanger sequencing.

Modified classification of pathogenicity
In accordance with the American College of Medical Genetics (ACMG) criteria, 19 in a first step, only pathogenic or likely pathogenic variants were selected consistent with a very high probability of molecular diagnostic confirmation (Table 1).Subsequently, variants with formally unclear clinical significance (VUS) in the core genes were reassessed (Table 2).For this purpose, five additional objective criteria were established to cover variant features that support the pathogenicity of a variant but are not represented in the actual ACMG guidelines so far.Two novel criteria were classified as moderate and three as supporting (Supplementary Table 2).Additionally, we established the term of VUS+, defined as variants that do not meet the original criteria for likely pathogenic or pathogenic variants but fulfil at least one of the new criteria supporting their pathogenicity.

Additional methods
Detailed information about long-read next-generation sequencing (NGS), sphingolipid profiling and electrophysiology is provided in the Supplementary material.

Results
The cohort studied was composed of patients who presented at the participating centres or whose findings were referred from peripheral hospitals or treating physicians.The inclusion of patients from countries or regions with limited resources led to variable availability of clinical information (Supplementary Fig. 1).Inclusion criteria for the genetic test were the suspicion of CIP/HSAN due to a decreasing or absent pain sensation and written consent to participate in the study.Cases were excluded if a secondary cause of insensitivity to pain such as leprosy or abusive injury was confirmed or suspected.The predefined inclusion and exclusion criteria are detailed in the 'Materials and methods' section.A cohort of 78 patients from 73 families with the suspected diagnosis of CIP/HSAN had been analysed by NGS using gene panels, whole exome (WES) or whole genome sequencing (WGS).The core genes included in the analysis were the following 22 genes: ATL1, ATL3, DST, ELP1, GLA, KIF1A, NGF, NTRK1, PRDM12, RAB7A, RETREG1/ FAM134B, SCN9A, SCN11A, SPTLC1, SPTLC2, TTR, WNK1, MPV17, NAGLU, CLTCL1, FAAHP1 and FLVCR1.
By applying the ACMG criteria and our additional criteria for pathogenicity, 80 novel CIP/HSAN-related variants were identified in 78 patients (Tables 1 and 2 and Supplementary Table 3).Complete clinical and genetic details are given in Supplementary Table 1 and clinical images for a subset of patients are shown in Fig. 1.In some patients with recessive conditions, the novel variant occurred in compound heterozygosity with a second, previously described pathogenic or likely pathogenic variant as indicated in Tables 1 and 2. The mutation spectrum across all genes included missense variants, non-frameshift variants, frameshift variants and stop-gains whereby for some genes, exclusively one mutation type was found (e.g.missense variants in SPTLC1 and SPTLC2) (Fig. 2).For most genes, no mutational hot spot was identified and the variants showed a distribution pattern across the entire coding regions, for some genes also including adjacent splice sites.An exception was the WNK1 gene, for which a known clustering of variants was confirmed in the HSN2 exon known to code for part of the neuron-specific isoform.One of these variants, however, affected the pan-isoform of WNK1 in compound heterozygosity with an HSN2-specific variant.
Novel CIP/HSAN-related variants were identified in 12 different CIP/HSAN genes, namely in ) and FLVCR1 (n = 6) (Supplementary Fig. 2), where 'n' corresponds to the number of patients per gene.
Genomic data revealed a larger intragenic deletion of 1.3 kb (Patient 35) in NTRK1 and two larger intragenic deletions of 3.8 kb (Patient 60) and 3.4 kb (Patient 61) in SCN9A, respectively.Patient 62 showed a 322 kb deletion spanning the entire SCN9A locus.In these patients, the deletion was in a compound  heterozygous situation with a single nucleotide variant on the other allele.For the 3.4 kb deletion (Patient 61) spanning exon 20 of SCN9A, the DNA quality was sufficient to determine the exact size of the maternally inherited deletion by long-read sequencing (Oxford Nanopore Technologies, ONT) (Supplementary Fig. 3A).
The deletion had a size of 3421 bp [chr2:166,235,764-166,239,185 (hg38)] and was further confirmed by quantitative PCR (qPCR) (Supplementary Fig. 3B).The father is a carrier of the single base pair deletion c.5318del, confirming compound heterozygosity in the index patient.For eight patients carrying SPTLC1 and SPTLC2 variants, plasma samples were available and 1-deoxysphingolipids (1-deoxySL) levels were elevated in line with assumed pathogenicity of the variant (Patients 63-68 and Patients 71-72) (Supplementary Fig. 4).For the homozygous missense variant p.(Leu172Arg) in Na v 1.7 (SCN9A) in transmembrane segment 2 of channel domain I (D1), in silico predictions regarding a role in CIP were inconsistent, so functional studies were performed (Patient 42).The respective variant was electrophysiologically analysed upon heterologous expression in HEK293 cells and showed a complete loss-of-function in line with pathogenicity (Supplementary Fig. 5).

Discussion
This retrospective cross-sectional study was aimed at molecular characterization of patients with CIP/HSAN and to our knowledge, includes the largest number of molecularly resolved cases to date.
The study has its main limitation in that in some cases the inclusion criteria were met, but detailed clinical data were not available.This was because data were collected from multiple sites, some with only limited clinical research and documentation capabilities.This is a frequently observed difficulty in ultrarare diseases, as  Novel genetic variants causing HSAN/CIP BRAIN 2023: 146; 4880-4890 | 4885 access to patients is already a major hurdle.In addition, the total number of cases investigated at each centre over the years is variable, making it difficult to accurately determine detection rates for CIP/HSAN.The strength and focus of this study were therefore on molecular characterization and careful assessment of the pathogenicity of variants in CIP/HSAN-associated genes.Thus, a large cohort of patients with these extremely rare diseases could be studied in a collaborative network and patients from normally under-represented countries could also be included.
Overall, the study supports that SCN9A and NTRK1 are the most frequently mutated genes in congenital painlessness.Whereas CIP/ HSAN-related pathogenic SCN9A variants frequently lead to anosmia as a secondary symptom, NTRK1-related neuropathy is accompanied by lack of sweat gland innervation with anhidrosis and sometimes life-threatening hyperthermia.Intellectual disability was not observed in SCN9A-associated neuropathy and was present with a variable degree in NTRK1-related disease.In adult onset HSN/HSAN, pathogenic variants are most frequently found in the enzymes of the sphingolipid metabolism pathway (SPTLC1/2).The study has a bias at this point: previously, more pathogenic variants were described for SPTLC1 than for SPTLC2.We report more cases with SPTLC2 variants in this study, but this is due to the fact that  we solely report new disease-relevant variants.In the UK, for example, there is a high frequency of SPTLC1-related patients due to the p.(Cys133Trp) founder mutation.
In addition, to address these more frequently mutated genes, we were able to substantiate the role of genes that have so far only very rarely been described as the cause of CIP/HSAN.To date, only three causal variants have been reported in NGF, two of which are missense variants. 10,20,21The study expands the mutation spectrum to include two homozygous loss-of-function variants and further corroborates a clinical presentation broadly equivalent to that of pathogenic variants in the NGF receptor encoding gene, NTRK1.
For ATL3, only two causal missense variants have been described to date. 6,22As for the previous changes, the here identified heterozygous variants are missense changes located at very highly conserved residues of the protein.The data further support that a dominant-negative effect of missense variants is likely the central mechanism of ATL3-associated disease.A recent report of an early stop-gain variant in ATL3, p.(Arg6Ter) 23 would argue against this assumption, but proof of pathogenicity of this variant is pending.FLVCR1 variants have predominantly been reported in cases of autosomal-recessive posterior column ataxia with retinitis pigmentosa (PCARP). 246][27] The findings in this study based on six additional patients show that the clinical transitions between PCARP and HSAN are fluid and often result in a complex phenotype with overlapping symptoms.
In several patients, the underlying genetic variants are immediately classifiable as likely pathogenic or pathogenic according to ACMG criteria.A greater difficulty arises with missense variants that often have to be classified as VUS.Here, further parameters such as homozygosity/compound heterozygosity with a likely pathogenic or very rare variant for recessive disorders, occurrence in more than one affected individual, a highly specific phenotype, detailed evaluation of the functionally critical domains and amino acids of the protein, or the lack of evidence of another genetic cause by broad genetic screening using WES or WGS were used to prove the pathogenicity of the suspicious variants.For all variants listed in this study, those that were formally classified as VUS according to ACMG criteria were considered as VUS+ (i.e.assumed to be likely pathogenic) or likely pathogenic based on such additional criteria.This critical review of variants was performed to exclude false positives as much as possible and to provide reliable genetic counselling to patients and their families.
Further functional assessment of VUS in SPTLC1 or SPTLC2 was achieved by detecting toxic sphingolipid species in the serum of patients by mass spectrometry.Both genes encode key enzymes of the de novo sphingolipid synthesis pathway, the so-called serine palmitoyltransferases (SPT).Pathogenic gain-of-function variants in SPTLC1 and SPTLC2 lead to the increased formation of toxic 1-deoxySL, which have been measured in patient's plasma to further assess suspicious VUS.Heterologous expression studies with functional measures were additionally performed in selected cases to corroborate pathogenicity of a VUS.As an additional example, we show whole-cell voltage-clamp recordings of HEK293 cells transiently expressing the SCN9A missense variant p.(Leu172Arg), confirming a complete loss-of-function of this variant.
The analysis of larger deletions has long been a difficulty in NGS-based diagnostics, but with increasingly better bioinformatic algorithms and new sequencing methods, this type of genetic alteration can more frequently be detected in CIP/HSAN.In rare cases, large deletions in NTRK1 have already been reported in HSAN. 28ur results show another case of an NTRK1 deletion and, in addition, three patients with larger deletions in SCN9A confirming their expected relevance in CIP/HSAN.Determination of copy number variants (CNV) from genomic data is therefore generally recommended in the case of an underlying loss-of-function mechanism.Long-read sequencing technologies, such as nanopore sequencing, have proven useful for rapidly determining the size and position of deletions with base pair precision, 29 as we also exemplify in a case of a 3.4 kb SCN9A deletion.
Another feature in CIP/HSAN concerns isoform-specific pathogenic variants.HSAN-relevant recessive variants in WNK1 cluster in a neuron-specific alternatively spliced exon (HSN2 exon) of the gene, whereas biallelic pan-WNK1 loss is most likely lethal.We report here one of the rare cases of a compound heterozygosity for a mutation in the neuron-specific exon in trans with a loss-of-function mutation affecting the pan-isoform of WNK1, similarly to a previous report. 30n conclusion, our results broaden the mutational spectrum of CIP/HSAN and the cohort provides a framework for natural history studies and improvement of care in these rare debilitating conditions.
compound heterozygosity, confirmed by segregation analyses; F/M = fractures and/or mutilations; het = heterozygous; hom = homozygosity, confirmed by segregation analyses; hom° = homozygosity, but no parental samples available for segregation analyses; Inh.= inheritance; ID = intellectual disability; LPV = likely pathogenic variant; Mo = motor dysfunction; red = reduced; PP = pain perception; PV = pathogenic variant; SL = skin lesions (including ulcerations).*Suspected clinical diagnosis of HSAN, no further clinical information available.For clinical data, '+' indicates the presence and '-' the absence of symptoms in the respective category.

Figure 2
Figure2Novel variants in known CIP/HSAN genes.The location within the respective genes is shown for all variants identified in this study.Green = missense/non-frameshift; yellow = splicing; red = truncating.For WNK1, the neuron-specific exon HSN2 is highlighted, in which the majority of the known pathogenic variants (to date) is located.