Clinical characteristics and genetic spectrum of 26 individuals of Chinese origin with primary ciliary dyskinesia

Motile cilia and flagella are highly conserved organelles that have a well-organized ‘9 + 2’ axoneme—9 peripheral microtubule doublets and a central pair of microtubule doublets—associated with multiprotein ultrastructure components, such as outer dynein arms, inner dynein arms, nexin-dynein regulatory complexes, and radial spokes rhythmically bending and beating in a proper pattern to expel the flow of extracellular fluid [1, 2]. Primary ciliary dyskinesia (PCD, OMIM ID: 244400), which is characterized by the impaired function of motile cilia, is a rare, genetically heterogeneous, autosomal recessive or X-linked disorder with an estimated incidence of 1:10,000–1:20,000 [3]. PCD has a wide range of clinical manifestations, including neonatal respiratory distress, year-round cough productive of sputum, persistent otitis media, bronchiectasis, infertility, and left–right laterality defects [4]. A specific subset of PCD, also known as Kartagener Syndrome (KS), is characterized by bronchiectasis, nasosinusitis and situs inversus [2].

As our understanding of genetics and clinical diagnosis of PCD far improves, multiple pathogenic genes of PCD have been reported, most of which encode axonemal, cytoplasmic, and regulatory proteins involved in the assembly, structure, and function of motile cilia, and the number of causative genes identified to date has significantly increased to 51 [5]. Specific PCD pathogenic genes cause distinct clinical symptoms and disease severities, making PCD a highly heterogenous disease and contributing to challenges for definitive diagnosis [4]. Some pathogenic genes lead to specific ultrastructural defects in the ciliary axonemal structure, which can be diagnosed by transmission electron microscopy (TEM) [1, 4, 6]. Therefore, TEM was a gold standard diagnostic test for PCD until the finding that some genes, such as DNAH11, CCDC65, and DRC1, give rise to PCD symptoms with normal cross-sections of the motile cilia axoneme [7‐9]. It is predicted that motile ciliary cross section defects can be detected in 70% of PCD cases [6, 10]. It is hard to diagnose PCD for there is no definitive definition of PCD and no strictly agreed criteria for PCD testing. The diagnosis of PCD may require combinations of copious tests, including nasal nitric oxide (nNO) measurement, digital high-speed video microscopy with ciliary beat pattern analysis (HSVA), immunofluorescence (IF) imaging for specific axonemal proteins, and gene testing [4, 11, 12]. Internationally, nNO measurement is 98% sensitive and 99% specific (cut-off value is 77 nL/min) for identifying individuals with PCD [2]. However, nNO measurement has only been available in China in the past 5 years, and the only related research showed that the best sensitivity is 86.1% and specificity is 91.4% (cut-off value is 76 nL/min) [13]. Given the lack of an agreed standard for nNO measurement or optimal cut-off values in a Chinese population, nNO measurement may suggest the presence of PCD but cannot be considered diagnostic in China. PICADAR, which is a score, contains characteristics of daily wet cough that started in early childhood, chest symptoms in the neonatal period, neonatal unit experience, situs abnormality, congenital heart defect, persistent perennial rhinitis, and chronic ear or hearing symptoms; and represents a simple diagnostic clinical prediction rule with good accuracy and validity for PCD [14]. Nevertheless, some of our patients were either older at presentation, or were born at home without significant medical contact in the neonatal period, meaning the use of PICADAR would not have been applicable. HSVA had excellent 96% sensitivity and 91% specificity to diagnose PCD [15]. However, methods for HSVA including sampling techniques, microscopes, cameras, temperature during analysis, software and evaluation criteria differ among centers [3]. A lack of standardized protocol restricts the application of HSVA in PCD diagnosis in isolation. In previous study, Immunofluorescence successfully identified 22 of 25 patients with PCD and normal staining in all 252 who were highly unlikely to be considered PCD, showing a high specificity but a limited sensitivity [16]. There is no single gold standard for PCD diagnosis so far [4, 10].

As the number of pathogenic genes underlying PCD rapidly increases to over 50 to date, the use of a genetic panel is hindered due to disadvantages, such as the limited gene-detection number, slow gene-updating speed, high renewing price, and unavailability to unknown candidate genes. A recent study demonstrated only a 67.6% detection rate for PCD when using a genetic panel targeted for 26 known PCD relevant genes and 284 additional candidate genes [17]. Compared with genetic panel, WES is more expensive but is associated with in an improved diagnostic rate (1.25 times of genetic panel's diagnostic yield) [18]. Meanwhile, the cost of WES and WGS has been decreased rapidly [18, 19]. In our cohort, the cost of WES and low-pass WGS of one patient is 1500RMB and 2100RMB, respectively. Hence, WES combined with low-pass WGS is our best choice to rapidly identify pathogenic/likely pathogenic single nucleotide variations (SNVs) and copy number variations (CNVs) in expanded candidate genes.

In this study, we demonstrated the practical PCD diagnostic criteria, emphasized the use of WES and low-pass WGS in combination to detect putative deleterious variations in 51 known PCD-causing genes and 42 suspected pathogenic genes, and supported the cost-saving potential of diagnostic low-pass WGS.

During recent years, more and more genes relevant to PCD have been identified due to our further comprehension of PCD and decreasing price for next generation sequencing. The number of known PCD-causative genes has grown to 51, and this number is still growing. PCD genetic panels with a limited number of genes tested are no longer practical for PCD diagnosis, and other recently developed methods to diagnose PCD, such as nNO, HSVA and three-dimensional structured illumination microscopy or stochastic optical reconstruction microscopy for PCD protein dysfunction assessment, can be expensive, unavailable, or difficult to standardize [11, 12]. Here, we used a process combining clinical symptoms and WES accompanied by low-pass WGS, which provides a unbiased portrayal of potentially causative variants and helps reveal CNVs that are not sensitive to WES, to help confirm PCD diagnosis. This diagnostic approach was an alternative and convenient diagnostic way for PCD patients who were reluctant to undergo invasive procedures such as bronchoscopy to obtain tissue for HSVA, IF, TEM. In our study, the mutation detection rate was as high as 73.1%, which slightly exceeds the previous 70% genetic detection rate [6]. This practical molecular genetic technique was also demonstrated to reveal the mutation characteristics of 76% of patients with PCD in a study by Marshall [22]. WES in combination with low-pass WGS has become a vital auxiliary tool for clinicians regarding disease diagnosis, decision-making and genetic precaution. For some patients with uncertain diagnoses, genetic screening can enable earlier recognition of PCD and consequently, appropriate treatment which will benefit patients and improve their quality of life. Even for those patients with typical PCD phenotypes such as KS, application of WES/WGS is still necessary as an auxiliary confirmation, for KS alone has not been classified as a standard criterion of PCD confirmation [4]. What’s more, WES/WGS can help to identify novel mutations, facilitate diagnosis of additional PCD cases with the same variants, help future prenatal genetic consulting and reveal hotspots for future gene therapy. New gene therapy approaches have been applied using lentiviral transduction and transcription activator-like effector nucleases (TALEN) technology for restoration of gene expression in vitro during recent years [23]. High-frequency mutation c.8383C > T (p.R2795X) in DNAH5 could be a hotspot and candidate target for gene therapy in Chinese PCD patients. However, we should also keep in mind that genetic screening cannot fully supplant specialized center diagnosis so far. Genetic testing has its limitation in poorly described gene variants, and genetic causes of about one-third PCD patients still remain unsolved [1].

PCD is a highly heterogeneous disease whose diagnosis can be obscure, and the age at onset is difficult to ascertain partly because of its atypical clinical manifestations and nonawareness of the disorder for most of the physicians in basic hospitals in China. In this study, the median age at diagnosis of 24.5 years was relatively high in all patients with suspected PCD, which may have mostly been due to selection bias because PUMCH is a hospital for adults. Selection bias also explains the difference in the age at diagnosis between our study and two other recently published studies on children with PCD in China [24, 25]. There were no other major differences between Chinese children and adults with PCD in any aspects of clinical presentations or genetics. Our study prompts healthcare providers of adults to consider PCD as a differential diagnosis for patients with related symptoms, especially in basic hospitals where physicians are not well-trained to diagnose rare disorders.

In our study, patients with PCD were confirmed based on Kartagener syndrome (heterotaxis phenotypes), ciliary axonemal defects by TEM, and genetic tests of compound heterozygous or homozygous mutations in known PCD-causing genes. After this confirming procedure, the median age at onset was 1.5 years, which is in accordance with the early onset feature of PCD. Compared to that in a previous study, the prevalence of cough and/or productive cough in this study was similar, but the prevalence of otitis media (13.0%) was significantly less than that in previous studies (weighted mean 74%) [26]. This finding may be because some of our patients forgot their childhood history of otitis media, since they came to PUMCH in their later years. In addition, it is more difficult to suspect PCD in patients who show only otitis media without other obvious respiratory symptoms. The high prevalence of bronchiectasis in our study (95.8%) compared with that in a previous study (weighted mean 56%) suggested that patients with PCD would be more likely to develop bronchiectasis at an older age [26]. Computed tomography showed the predominance of bronchiectasis in the middle and lower lobes, which similarly occurred in a previous study [27].

Our study helps reveal the mutation spectrum and genotype–phenotype relationship in Chinese patients with PCD. Among all candidate PCD-relevant genes, DNAH5 was the most prevalent disease-causing gene, occurring in 23.1% of all patients with PCD, compared to a previous study by Failly in which only 15% of 89 unrelated individuals of European origin with PCD were associated with DNAH5 and in a study by Hornef in which 28% of 109 individuals of European or North American origin with PCD carried mutations clustering within exons 34, 50, 63, 76, and 77, and no c.8383C > T variants were observed [20, 28]. In contrast, the c.8383C > T stop-gain mutation in DNAH5 was a hotspot among patients of Chinese origin with PCD in our study. It is said that DNAI1, the initial PCD-implicated gene, occurs in 9% of all patients with identified PCD [6]. To our surprise, no patient had homozygous or compound heterozygous variations in DNAI1 in our cohort study. CCDC39 and CCDC40 are the most prevalent mutated genes in individuals of Egyptian origin with PCD [29]. A comparable phenomenon was observed in 58 Tunisian patients with PCD in whom CCDC39 and c.2190del in CCDC39 was a mutation hotspot, whereas deleterious mutations in CCDC39 or CCDC40 were rarely observed in Chinese patients with PCD and only one case involved CCDC40 dysfunction [30]. These results elucidate the disparate genetic spectra of patients of Chinese and other ethnic backgrounds with PCD.

NGS provides us with abundant novel variants without bias as well as a great challenge to determine pathogenic mutations in individuals with PCD. In this cohort, 8 of 32 variations found in our patients were marked as damaging in the HGMD with definite pathogenicity. A total of 24 novel mutations were found, most of which were nonsense, premRNA splicing, and frameshift mutations, and presumably had deleterious effects. The remaining missense variants were either pathogenic/likely pathogenic mutations assessed by ACMG/AMP or VUS with a high damaging score provided by VarCards. As described in the results, VUS, assessed by ACMG standards, could be deemed to be deleterious by VarCards with relatively high damaging scores because of either an extremely low MAF of less than 0.0001 (c.9631C > T (p.P3211S) in DNAH11) or location in a relatively conserved domain of protein (c.53T > C (p.I18T) mutation in LRRC6), which further provided evidence for their pathogenicity. A more complicated case was patient 23, who carried a c.4596 + 1G > A mutation in DNAH5 of paternal origin and two missense variants, c.7520G > C (p.W2507S) and c.5518C > T (p.R1840W) of maternal origin. Both missense variations were assumed to be VUS by ACMG but deleterious by VarCards, with c.5518C > T slightly more detrimental (damaging score of 0.61) than c.7520G > C (damaging score of 0.57). However, the possibility of the synergetic effects of both variants cannot be ruled out. Overall, identified variants were classified as disease-causing based on their minor allele frequencies, biological relevance to the disease, bioinformatic tools applied, and previous results published. However, the true effects of those putative deleterious mutations are uncertain, and more functional studies or long-term follow-ups are warranted.

In terms of the relationship between genotypes and axonemal defects in patients with PCD, two cases with mutated CCNO had absent or short cilia, consistent with the results of a previous study by Wallmeier that CCNO is involved in cilia generation [31]. Patient 15 with pathogenic variations in DNAH1 had inner dynein defects. This finding can be explained by the DNAH1 gene function of encoding axonemal inner dynein arms [32]. Dysfunction of DNAH5 regularly caused outer dynein defects in the ciliary axoneme in a previous report [20]. In our study, however, abnormalities in the axonemal structure were not detected in patient 14, who had deleterious variations in DNAH5, perhaps due to inappropriate sampling sites for TEM. In addition, DNAH11 was thought not to cause any axonemal defects, whereas in patient 25 with compound heterozygous mutations in DNAH11, ciliary absence in respiratory epithelial cells was detected, which may also have been due to the sites of sampling and may require repeating TEM examinations of the airway epithelium [33].