Phenotype‐Genotype Correlations in ABCA3 Patients—The RespiRare Cohort

ABSTRACT

Background

ATP‐binding cassette transporter A3 (ABCA3) deficiency is one of the most severe causes of childhood interstitial lung diseases (chILD). This study aims to report the RespiRare ABCA3 cohort and to establish phenotype‐genotype correlations.

Methods

Phenotypic and genotypic data of patients under 18 years were retrospectively included (1995–2023) in the RespiRare centers. The initial presentation and evolution of the subjects was analyzed depending on their genotype.

Results

The ABCA3 cohort comprised 36 children (30 families), including 5.5%, 22%, and 72% of null/null (no protein), null/other (potential residual function) and other/other genotypes respectively. A neonatal respiratory distress syndrome was observed in 31 (86%) subjects and 27 (75%) died at a median age of 3 months. The 5‐year overall survival was 25% with an overall median survival of 0.33 year (IQR 0.09–4.43). A neonatal onset (p = 0.009) and the presence of pulmonary hypertension (p = 0.037) impaired the prognosis. At the last follow‐up, the survival rates were 0/2 (0%), 4/8 (50%) and 6/26 (23%) in the null/null, null/other and other/other groups respectively. Eight of the 12 subjects who survived beyond 1 year carried at least one missense variant outside the nucleotide‐binding domains (NBD) (n = 9) or the hypomorphic p.(Glu292Val) variant (n = 1).

Conclusion

The variable presentation and outcome of chILD due to ABCA3 pathogenic variants are linked to the underlying genotype. Neonatal onset, null variants, and variants involving the NBD are of peculiar severity.

Abbreviations

ABC

ATP‐binding cassette

BPD

Bronchopulmonary dysplasia

chILD

childhood interstitial lung disease

ECD

Extracytosolic domain

ECMO

Extracorporeal membrane oxygenation

ETI

Elexacaftor/Tezacaftor/Ivacaftor

GGO

Ground glass opacifications

ILD

Interstitial lung disease

IQR

Interquartile range

MDD

Multidisciplinary discussion

NBD

Nucleotide binding domain

PHT

Pulmonary hypertension

RDS

Respiratory distress syndrome

1. Introduction

Childhood interstitial lung diseases (chILD) represent a heterogeneous group of rare and severe diseases [1]. Among the 10%–20% of genetic causes, ATP‐binding cassette transporter 3 (ABCA3) deficiency, a recessive disease, is one of the most severe causes of chILD [2, 3].

ABCA3 is a member of a large family of proteins that hydrolyze ATP to transport substrate across membranes. Expressed by alveolar epithelial cells, it transports phospholipids into lipid lamellar bodies where surfactant proteins (SP)‐B and SP‐C are processed [4]. ABCA3 (NM_001089), an 80 kb gene located on chromosome 16 encodes a 1704 amino acid protein. More than 300 pathogenic variants of ABCA3 have been described, the p.(Glu292Val), E292V, variant, being the most frequent [5, 6]. Classifications of ABCA3 variants have been proposed over time, usually based on in vitro functional studies and on the expected functional defect of the protein [7, 8, 9].

Bi‐allelic variants in ABCA3 were first associated with neonatal and infant forms of severe ILD [7]. Cohorts of ABCA3‐related ILD reveal substantial phenotypic heterogeneity including adult forms of the disease (Table 1) [8, 10, 11, 12, 13, 14, 15, 16].

Table 1.

ABCA3 deficiency cohorts.

Country or area	Number of subjects	Pediatric/adult	Reference
France	7	Pediatric	Flamein et al. Hum Mol genet [8]
USA	185	Pediatric	Wambach et al. AJRCCM [9]
Argentina	7	Pediatric	Balinotti et al. Ped Pulm 2023 [16]
Germany and Europe	44	Pediatric	Li et al. Thorax 2023 [11]
France	14	Adult (including five subjects with pediatric onset)	Diesler et al. Respirology [10]

The therapeutic management of surfactant metabolism disorders currently involves nonspecific therapies such as corticosteroids, azithromycin, or hydroxychloroquine [2, 17]. Recently, cyclosporin A has also shown some in vitro efficacy and cystic fibrosis transmembrane conductance regulator (CFTR) modulators have also been suggested as a promising therapeutic option, at least for some ABCA3 genotypes [18, 19, 20, 21].

The aim of this study was to describe phenotype‐genotype correlations and prognosis factors of patients with ABCA3 deficiency in the RespiRare network, that could guide management and genetic counseling.

2. Methods

2.1. Subjects

The RespiRare centers recruited subjects under 18 years at diagnosis of ILD between 1995 and 2023 who carry two ABCA3 variants (homozygous or compound heterozygous).

2.1.1. Data Collection

The data were collected retrospectively from the French RespiRare database, the national multidisciplinary discussions (MDD) for chILD, and the subject records, with no overlap with European published cohorts [22]. The following information was retrieved: gender, gestational age, family history of ILD, and at first symptoms: age, anthropometric measures, symptoms and Fan severity score, chest X‐ray, CT‐scan, bronchoscopy and bronchoalveolar lavage, echocardiography, histopathology of lung biopsies, genetic analyses and therapeutic management and outcome at last follow‐up. The Fan score, ranging from 1 to 5 points was proposed by Leland L. Fan in 1997 to easily assess chILD severity [23].

2.1.2. Genetic Analyses

The identified variants were classified according to the five‐class system of the American College of Medical Genetics and Genomics (ACMG) using in silico tools, and published functional tests when available for previously described variants: pathogenic, likely pathogenic, variant of uncertain significance, likely benign, and benign [24]. The SpliceAI prediction tool was used to assess the potential effect on splicing of all variants, with a delta score threshold of 0.2. Nonsense and frameshift variants leading to a premature stop codon located upstream the 55 nucleotides of the last exon–exon junction were considered to induce nonsense‐mediated mRNA decay (NMD) [25]. Only pathogenic and likely pathogenic variants were collected from Sanger or Next Generation Sequencing (NGS). When available, parental DNA was analyzed for confirmation of compound heterozygosity in the proband. The following ABCA3 variants were classified as “null”: nonsense and frameshift variants resulting in a premature stop codon predicted to induce NMD, variants that were predicted to fully disrupt a splice site, and deletions of at least an entire exon leading to a frameshift, all of which are predicted to result in an absence of protein. The “other” variants (also referred to as “hypo”) included missense variants, moderate splice‐site variants, in‐frame insertions or deletions, and distal nonsense variants, all predicted to alter the protein function. The domain of missense variants was also taken into account.

2.1.3. Ethics

Parents or guardians gave informed consent for participation in the study. The study was approved by the local ethical evaluation committee of the French Society of Pulmonology (Société de Pneumologie de Langue Française SPLF CEPRO, reference n°2022‐054). The data collection was validated by the regulatory authority (Comité National de l'Informatique et des Libertés, CNIL n°08.015bis) and by the ethics committee (Comité Consultatif sur le Traitement de l'Information en matière de Recherche dans le domaine de la Santé, CCTIRS, 20080328).

2.1.4. Statistical Analyses

Prism (GraphPad) was used for statistical analysis. Qualitative/categorical variables were presented as numbers and percentages. Quantitative/continuous variables were presented as median and interquartile range (IQR). Chi‐square and Fisher's exact tests were used to compare nominal variables, and Kruskal‐Wallis or Mann‐Whitney tests were used to compare ordinal variables. A p‐value ≤ 0.05 was considered significant. Kaplan‐Meier curves were used for survival analysis and compared using the log‐rank test.

3. Results

Between 1995 and 2023, 36 subjects with pathogenic variants in ABCA3 were identified within 30 families (10 homozygosity and 20 compound heterozygosity). A majority of them were retrieved from the chILD RespiRare cohort (2000–2023) and 5 additional subjects were included within the 1995–2000 and February to December 2023 periods (Table S1). Among the 36 subjects, 26 (72%) had an other/other genotype, 8 (22%) had a null/other genotype and only two subjects (5.5%) had a null/null genotype (Table 2 and Table S2). The probands were of diverse ancestries, with no gender predominance. The majority of the newborns were term‐born (n = 34, 94%) with normal measurements at birth. There were no extreme preterm births, only late prematurity. In all subjects, 30% were familial cases (Table 2). Consanguinity was reported in 33% of the families (Table S2).

Table 2.

Characteristics of subjects with ABCA3 pathogenic variants.

	Total	null/null	null/other	other/other	p value
Sample size, n (%)	36	2 (5.5%)	8 (22%)	26 (72%)
Initial presentation
Description of the population, n (%)
Male, n (%)	15 (42%)	0	3 (37%)	12 (46%)	0.427
Term (≥ 37WG), n (%)	34 (94%)	2	7 (87%)	25 (96%)	0.202
Neonatal RDS, n (%)	31 (86%)	2	6 (75%)	23 (88%)	0.530
Family history of ILD, n (%)	15 (42%)	0	3 (37%)	12(46%)	0.427
Consanguinity, n (%)	15 (42%)	0	3 (37%)	12 (46%)	0.279
Neonatal respiratory support, n (%)					0.568
None, n (%)	5 (14%)	0	2 (25%)	3 (12%)
Oxygen only, n (%)	1 (3%)	0	0 (0%)	1 (4%)
Non invasive ventilation only, n (%)	4 (11%)	0	1 (12%)	3 (12%)
Mechanical ventilation, n (%)	26 (72%)	2	6 (75%)	19 (73%)
Initial Fan score, n (%)					0.610
3/5	5 (14%)	0	2 (25%)	3 (12%)
4/5	20 (55%)	2	5 (62%)	14 (54%)
5/5	11 (30%)	0	2 (25%)	9 (35%)
Histological analysis
Not performed n (%)	25 (69%)	2	7 (87%)	17 (65%)
Biopsy n (%)	10 (28%)	0	2 (25%)	8 (31%)
Autopsy n (%)	1 (3%)	0	0 (0%)	1 (4%)
Outcome
Total death (%)	27 (75%)	2 (100%)	5 (62%)	20 (77%)	0.500
Age at death
< 1 year, n (%)	22 (61%)	1	4 (50%)	18 (69%)	0.553
1–5 year, n (%)	3 (8%)	1	1 (12%)	1 (4%)
Including transplanted, n (%)	1 (3%)	0	0 (0%)	1 (4%)
> 5 year, n (%)	1 (3%)	0	0 (0%)	1 (4%)
Age at death					0.632
Mean (years, SD)	1.21 (3.50)	0.94 (1.08)	1.14 (2.16)	1.25 (3.96)
Median (years, IQR)	0.25 (0.08–0.42)	0.94 (0.55–1.32)	0.17 (0.12–0.33)	0.25 (0.07–0.42)
Total alive at last follow‐up (%)	9 (25%)	0 (0%)	4 (50%)	6 (23%)	0.500
Age at last follow‐up
< 1 year, n (%)	1 (3%)	0	1 (12%)	0 (0%)	0.230
1–5 years, n (%)	1 (3%)	0	1 (12%)	0 (0%)
> 5 years, n (%)	7 (19%)	0	2 (25%)	6 (23%)

3.1. Clinical Presentation

Most subjects (n = 31, 86%) presented a respiratory distress syndrome (RDS) at birth, including the two subjects with a null/null genotype, while only 2 (5%) subjects showed symptoms in the first year of life, and 3 (8%) subjects in childhood (Figure 1a).

Figure 1.

Clinical and radiological presentations of subjects with ABCA3 pathogenic variants. (a) Age at onset of the symptoms. These data were compared by genotype groups using the chi‐square test (p = 0.8329). (b–f) Main patterns seen in axial high‐resolution CT (HRCT). (b–d) newborns subjects with null/null (SURF549), null/other (FR003SU660‐1159) and other/other (19GM00649) genotypes respectively. All of them presented dense and diffuse GGO (under mechanical ventilation). (e) Long‐term evolution of a subject (17GM00717) with an other/other genotype at 25 days, 4 years 10 months and 6 years 2 months, decrease of GGO intensity and extension over time; at 5 years, diffuse parenchymal destruction with emphysema. (f) Long‐term evolution of a subject (PPD507) with an other/other genotype at 8, 10 and 21‐year‐old subject with a chILD evolving since birth, multiple cystic images, diffuse mild GGO and intralobular septal thickening.

The presentation was predominantly severe with a 4 or 5/5 initial Fan score in 31 (86%) subjects [23]. Other symptoms reported at diagnosis were digital clubbing (n = 3/36), pectus excavatum (n = 1/36), pneumothorax (n = 3/36), growth retardation (n = 5/36) and gastrointestinal reflux (n = 4/36).

3.2. Chest X‐Ray and High‐Resolution CT Analysis

Chest x‐rays were all abnormal with ground glass opacities and/or emphysema and led to perform the chest CT scans. At least one high‐resolution CT scan was available for 30 subjects (83%), the remaining six subjects were too critically ill to undergo a CT imaging (Figure 1b‐f). Under 6 months, all subjects had diffuse or patchy ground‐glass opacities (GGO) involving multiple lobes, and this pattern persisted over time with a gradual development of septal thickening, cysts, parenchymal destruction with emphysema and architectural distortion. As observed in subjects with a long‐term evolution, and although not quantified, the intensity and the extent of the GGO appeared to gradually decrease over time.

3.3. Bronchoalveolar Lavage and Histology of the Lungs

Bronchoalveolar lavage (BAL) was performed in 17 subjects (47%) as part of the usual chILD work‐up. Cytological analysis (n = 11 subjects with available data), revealed a normal to increased total cell count with an elevated neutrophil count, indicating various degrees of inflammation.

Lung histology (n = 11, 30%), including 10 lung biopsies (28%) and one autopsy (Table 2) supported a diagnosis of surfactant metabolism disorder with thickened alveolar septa and hyperplastic alveolar epithelial cells in 6/11 (36%). Two subjects exhibited a pattern of diffuse developmental disorder in term infants. The results were unavailable for three subjects. In total, 4 (36%) had features of lung fibrosis. In two siblings with electron microscopy analyses (P23 and PPD507/D34, carrying the homozygous “other” p.(Asp253His) variant), lamellar bodies were small and dense [8].

3.4. Management

Only 5 (14%) subjects, who had null/other or other/other genotypes (identified in Table S2), did not require any respiratory support. One of them carried the hypomorphic E292V variant. Twenty‐six (72%) subjects required mechanical ventilation, while two subjects among the 11 (29%) with supra‐systemic pulmonary hypertension (PHT) required extracorporeal membrane oxygenation (ECMO). At birth, exogenous surfactant was administered intratracheally to 15 (42%) subjects with a positive but very transient clinical improvement and a decrease in the ventilatory parameters in half of them.

The most common therapy administered to subjects was corticosteroids in 27/28 (96%) of them, typically as monthly intravenous pulses (300 mg/m²/day for 3 days/month) with an initial addition of oral prednisolone. The addition of azithromycin and hydroxychloroquine was reported in 13/28 (46%) subjects. One infant underwent a lung transplantation at 30 months of age but unfortunately died 2 months later. Most of the subjects (n = 32, 89%) received enteral nutrition through via nasogastric tube (n = 26, 81%) or gastrostomy (n = 6, 19%).

3.5. Outcomes

The 5‐year overall survival was 25% with an overall median survival of 0.33 year (IQR 0.09–4.43). Only 9 (25%) were alive at last follow‐up (Figure 2 and Table 2) including one patient carrying the p.(Glu292Val) variant, two subjects carrying the homozygous p.(Asp253His) variant and two other subjects carrying the homozygous p.(Thr1173Arg) variant. The prognosis was correlated to the age at onset, the initial severity and the genotype. Subjects with a later onset (n = 5/36, 14%) had a better prognosis than those with neonatal onset (n = 31/36, 86%) (p = 0.009). Of the 11 subjects with a Fan score of 5, 10 (91%) died within the neonatal period. At the last follow‐up, the survival rates were 0/2 (0%), 4/8 (50%) and 6/26 (23%) in the null/null, null/other and other/other groups respectively, with a median survival of 11 months, 4 months and 4 months respectively.

Figure 2.

Kaplan‐Meier survival curve according to the ABCA3 genotype. The three survival curves according to genotype were compared using the log‐rank test (p‐value = 0.9417). [Color figure can be viewed at wileyonlinelibrary.com]

3.6. Distribution of Variants Across ABCA3

The study identified 43 distinct ABCA3 pathogenic variants distributed throughout the entire ABCA3 protein including 17 newly described (Table S3). Compound heterozygosity confirmation was obtained with parental DNA in 16/21 (76%) subjects. The majority were missense variants (n = 30) or predicted to result in a premature stop codon leading to NMD (n = 12), one variant was predicted to result in a longer protein, the frameshift p.(Leu1695Argfs*103) that results in a 1797 amino acid protein whereas the normal ABCA3 is 1704 amino acid long (Figure 3). The two main hotspots for the missense variants were the first extracytosolic domain (ECD1) (n = 9) and the second nucleotide binding domain (NBD2) (n = 5). Seven pathogenic variants were identified in more than one family (p.(Gly202Arg); p.(Arg208Trp); p.(Gly210Cys); p.(Asp253His); p.(Ala1027Pro); p.(Phe1077Ile); p.(Asp1539Asn)) (Table S3). Two of them were associated with a prolonged survival (over 2 years) in more than one family and may be hypomorphic variants: p.(Arg208Trp) (heterozygous), p.(Asp253His) (homozygous) [26]. Interestingly, all 11 (30%) subjects who survived beyond 2 years carried at least one allele with a missense variant outside the nucleotide‐binding domains (n = 9), the hypomorphic p.(Glu292Val) variant (n = 1) or the distal frameshift variant p.(Leu1695Argfs*103) (n = 1).

Figure 3.

ABCA3 transporter domain organization and localization of the pathogenic variants identified in the cohort. The missense pathogenic variants were primarily identified in the two extracytosolic domains (ECD1 and 2), and in the two nucleotide‐binding domains (NBD1 and 2). The boundaries of the helical transmembrane domains (TMD) were determined based on the 3D structure of the human protein (7W02 PDB crystal) and on TMHMM predictions. ECL, extracellular loop; IH, intracellular transverse helix; EH, external helix; RD, regulatory domain. Null variants resulting in a premature Stop codon are in purple. Created with Biorender. [Color figure can be viewed at wileyonlinelibrary.com]

4. Discussion

This study characterized a French cohort of 36 subjects carrying 43 distinct pathogenic ABCA3 variants, 17 being newly described. It confirms that a null/null genotype is associated with a fatal prognosis and suggests that subjects who survive beyond 1 year tend to carry missense variants located outside the NBDs.

In recent studies ABCA3 pathogenic variants account for 1.8%–4.4% of chILD and are to date the leading cause of chILD at birth [3, 27, 28, 29]. In this study, the data collection strategy (RespiRare database, MDD (17/36; 47% of the subjects), and subjects records) is probably close from exhaustiveness for ABCA3 identified cases in France [22, 30]. Only a few subjects with ILD benefited from a genetic analysis before 2000. The proportion has reached 543/706 (76.9%) in the recent RespiRare report [3]. However, the increasing availability of genetic tests in France now allows patients with ILD below 50 years to benefit from a NGS panel including surfactant metabolism genes. Although our data collection was as large as possible, undiagnosed cases may still exist due to the severity of the prognosis and the high rate of neonatal deaths. This may be particularly true for preterm infants, or conversely, for the less severe neonatal respiratory distress that spontaneously resolve [31].

The study confirms that the prognosis of chILD related to ABCA3 variants is extremely severe. As described, PHT was observed at presentation in a third of cases and was significantly associated with a poor prognosis [23]. The majority of subjects died within the first year of life (n = 22, 61%), with only 4 subjects reaching young adulthood. These results are in line with previous reports in which 44%–65% of subjects die within the neonatal period or before 1 year, with only a few reaching adulthood [9, 11, 28]. This poor prognosis is however genotype‐specific with null/null having a very poor prognosis and the other genotypes being more difficult to predict. Interestingly, for the 5 (13.8%) subjects who manifested after the neonatal period, a significant better prognosis was observed with 80% of these subjects being alive at follow‐up (p = 0.014), similarly to the report of Li et al. [11].

In subjects with ABCA3 variants who survived, the CT‐scan pattern varied depending on the age of the subjects. GGO was a constant feature but its extent and intensity decreased over time [8, 32, 33]. Conversely, fibrosing abnormalities appeared and extended, including cysts, parenchymal destruction with emphysema and distortion, and septal thickening.

ABCA3 deficiency is primarily a molecular diagnosis, and conclusive genetic results may obviate the need for lung biopsy to confirm the diagnosis. However, a limited number of subjects also underwent a lung biopsy or autopsy, mostly when genetic analyses were not widely available. Lung biopsy may also be considered for individuals with ABCA3 variants of uncertain significance or with other atypical clinical features. They revealed various patterns of chILD, most of which being consistent with a surfactant disorder or a diffuse developmental disorder. This later observation suggests that abnormal surfactant processing may affect lung development and impair the prognosis of ABCA3 patients [4, 7, 34]. Recently, genetic analyses have become quicker and lung biopsies are no longer commonly performed in ABCA3 patients.

The genetic analysis identified 43 pathogenic variants distributed throughout the entire ABCA3 gene. Two subjects with complex alleles were identified, confirming the importance of parental DNA testing for genotype confirmation [9]. Some phenotype‐genotype correlations have been highlighted or confirmed based on the type of variant, the location of the variant, and its hypomorphic features. We confirm that a null/null genotype, which corresponds to a complete ABCA3 deficiency due to loss‐of‐function variants, was associated with a fatal prognosis in infancy in 100% of cases [9]. In different studies, the null/null genotype was identified in up to 25.4% of subjects with an ABCA3‐deficiency [9, 11, 28].

“Other” missense variants that result in an altered protein function and/or protein conformation, may result in a potential residual ABCA3 activity. This probably explains the phenotypic variability in patients in terms of presentation and prognosis [35]. We underline that subjects who survive over 1 year old carry at least one variant that does not affect the NBD1 or NBD2. The NBD domains are located on the cytosolic side of the protein and comprise the Walker A domain which is crucial for ATP‐fixation in ABC proteins [36]. Therefore, its disruption is likely to be highly deleterious for ABCA3 function. Nevertheless, the localization of a variant outside the NBD domains does not necessarily preclude a highly deleterious effect (e.g., p.(Gly202Arg, p.(Pro246Leu)).

We also observed that the hypomorphic variant p.(Glu292Val), located in the second intracytoplasmic helix, appears to be associated with a better prognosis [6]. The proportion of heterozygous carriers of this variant is 1.15% in the European population, and up to 20 homozygous individuals are found in GnomAD v4.1 (a general population database) [37]. There is however compelling evidence that confirm its pathogenicity with functional analyses of this variant indicating reduced but not abolished ATP hydrolysis activity [19, 26, 38]. Regarding the clinical descriptions, one adult subjects carrying the homozygous p.(Glu292Val) variant was described in the literature, with a late presentation at 25 years and lung transplantation at 47 years following a long‐term evolution [15]. Other adult ABCA3 studies found that the p.(Glu292Val) variant accounted for up to 35.7% of the subjects with long‐term survival [10, 13, 15, 39, 40]. The evolution was more variable in pediatric subjects with chILD in whom the p.(Glu292Val) variant was identified in a homozygous state [5, 9, 41]. Similarly, the p.(Ser1262Gly) variant is also hypomorphic with 0.5% of carriers in the European population and 11 homozygous individuals in GnomAD [37, 38]. These hypomorphic variants may thus explain why, as in other series, the survival of the null/other group, that includes the two subjects with the p.(Glu292Val) and the subject with the p.(Ser1262Gly) variants, is higher than the survival of the other/other group [11].

In addition to the molecular variability, intra‐familial variability also exists. Two pairs of siblings carrying the “other” variants p.(Phe1077Ile) and p.(Asp253His) respectively showed a variable clinical expression [42]. In these two families, the second affected child showed a better prognosis than the first who died in the neonatal period. A better management at birth may have influenced the prognosis, but it could also be hypothesized that modifier genes or differential degrees of activation of intrinsic pathways (endoplasmic reticulum stress or cell death) or environmental triggers may also be involved [43]. Larger genetic screening by exome or genome sequencing may help answering this question.

The management of patients with ABCA3 deficiency is not consensual. Currently, corticosteroids represent the most common treatment for chILD supported by the European guidelines, especially in surfactant‐related disorders, as in the present study [2]. While individual experiences and case reports suggest their efficacy, this remains to be ascertained by international clinical trials. Furthermore, they appear to be insufficient in severe neonatal presentations. The present study did not report on anti‐fibrosing therapies. This promising therapeutic option was studied in chILD in an international clinical trial on nintedanib including participants aged 6–17 years [44]. Nevertheless, children below 6 years old, who represent the large majority of ABCA3 patients are not yet eligible. Although azithromycin and/or hydroxychloroquine are commonly used, their efficacy in chILD is only supported by a low degree of confidence [17, 42]. Recently, CFTR modulators have also been suggested as a potential promising treatment for ABCA3 variants. However, the efficacy of this approach is likely to vary depending on the ABCA3 genotype and the age at presentation. This assumption is based on the homology between CFTR (also called ABCC7) and ABCA3, both being members of the same ATP binding cassette family of proteins [45]. CFTR modulators have shown positive effects on maturation and function of some ABCA3 variants in vitro. These findings were recently confirmed through a compassionate program in three adults with ABCA3 deficiency in France with encouraging results [19, 20, 21]. Another promising therapeutic approach for severe forms of ABCA3 deficiency is neonatal gene therapy that has already demonstrated some degree of feasibility for SP‐B deficiency in vitro and in mice models [46, 47]. In the absence of these options, some patients may benefit from lung transplantation with a similar prognosis to other lung transplantations in children [48, 49, 50, 51].

In conclusion, this ABCA3 cohort study shows that the factors associated with a worse prognosis are neonatal presentation, presence of PHT, and null variants. Conversely, all the subjects who survived beyond 1 year carried at least one variant located outside the NBD domains. This information is critical for genetic counseling and individual prognosis and international studies are required to better understand phenotypic heterogeneity of this ultra‐rare disease.

Author Contributions

Manon Fleury: investigation, conceptualization, writing – original draft. Céline Delestrain: investigation, validation, writing – review and editing, resources. Alice Hadchouel: writing – review and editing, resources, validation. Julie Mazenq: writing – review and editing, resources, validation. Myriam Benhamida: validation, writing – review and editing, resources. Anne‐Sophie Bernard: validation, writing – review and editing, resources. Raphaël Borie: validation, writing – review and editing, resources. Jacques Brouard: validation, writing – review and editing, resources. Harriet Corvol: validation, writing – review and editing, resources. Pierrick Cros: writing – review and editing, validation, resources. Christophe Delacourt: validation, writing – review and editing, resources. Tifenn Desroziers: validation, writing – review and editing, resources. Jean‐Christophe Dubus: validation, writing – review and editing, resources. Carole Egron: validation, writing – review and editing, resources. Ralph Epaud: validation, writing – review and editing, resources. Michael Fayon: validation, writing – review and editing, resources. Aude Forgeron: validation, writing – review & editing, resources. Lisa Giovannini‐Chami: validation, writing – review and editing, resources. Christophe Marguet: validation, writing – review and editing, resources. Alexandra Masson‐Rouchaud: validation, writing – review and editing, resources. Hortense Petat: validation, writing – review and editing, resources. Marie‐Catherine Renoux: validation, writing – review and editing, resources. Léa Roditis: validation, writing – review and editing, Resources. Caroline Thumerelle: validation, writing – review and editing, resources. Clémentine Vigier: validation, writing – review and editing, resources. Aurore Coulomb L'Herminé: investigation, writing – review and editing, resources, validation, data curation. Hubert Ducou le Pointe: investigation, validation, writing – review and editing, resources, data curation. Pascale Fanen: writing – review and editing, validation, data curation, resources. Camille Fletcher: writing – review and editing, resources. Chiara Sileo: investigation, validation, writing – review and editing, data curation, resources. Laureline Berteloot: investigation, validation, writing – review and editing, resources. Camille Louvrier: investigation, writing – review and editing, software, data curation, resources, validation. Alix Becdelièvre: investigation, writing – review and editing, validation, software, data curation, resources. Marie Legendre: conceptualization, investigation, writing – original draft, writing – review and editing, validation, data curation, software, formal analysis, supervision, resources. Nadia Nathan: data curation, supervision, resources, software, formal analysis, project administration, validation, writing – review and editing, writing – original draft, investigation, conceptualization.

Ethics Statement

Parents or guardians gave informed consent for participation in the study. The study was approved by the local ethical evaluation committee of the French Society of Pulmonology (Société de Pneumologie de langue française SPLF CEPRO, reference n°2022‐054). The data collection was validated by the regulatory authority (Comité National de l'Informatique et des Libertés, CNIL n°08.015bis) and by the ethics committee (Comité consultatif sur le Traitement de l'Information en matière de Recherche dans le domaine de la Santé, CCTIRS, 20080328).

Conflicts of Interest

The authors have no conflict of interest to disclose in relation to this study. They have declared the following disclosures: Myriam Benhamida has declared Support for attending meetings and/or travel from the Société Francaise Mucoviscidose, Filière MUCO‐CFTR, VIATRIS. Raphaël Borie has declared consulting fees from Boerhinger Ingelheim, Ferrer, Sanofi; payment or honoraria for lectures, presentations, speakers bureaus, manuscript writing or educational events from Boerhinger Ingelheim and support for attending meetings and/or travel from Boerhinger Ingelheim. Harriet Corvol has declared being the head of the Alice Hadchouel has declared being the head of the Société de Pneumologie pédiatrique et allergologie (SP2A), unpaid. Ralph Epaud has declared consulting fees from Astra Zeneca; Payment or honoraria for lectures from GSK, Astra Zeneca, Menarini, support for attending meetings and/or travel from GSK, Astra Zeneca and participation on an advisory Board of Astra Zeneca and Novartis. Michaël Fayon has declared being the president of the SPOI (Indian Ocean Lung Society). Alice Hadchouel has declared being the head of the Scientific committee of the Société de Pneumologie pédiatrique et allergologie (SP2A), unpaid. Julie Mazenq has declared a PhD grand from Astra‐Zeneca. Nadia Nathan has declared grants from French Ministry of health, Million Dollar Bike Ride, Chancellerie des Universités, RespiFIL; payment for manuscript writing from La lettre du Pneumologue; and support for attending meetings from the ERS.

Supporting information

ABCA3 Supplemental R3 Sept10.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request. The datasets generated and/or analyzed during the current study are available in the supplemental material of this article and upon request. This includes raw data, processed data, and any additional materials necessary to replicate the study findings. For any further inquiries or requests for data, please contact the corresponding author.