This study was conducted with the approval of the Cincinnati Children's Hospital Medical Center institutional review board. All participants or their legal guardians gave written informed consent and minors gave assent. The pediatric participants with PAP underwent a complete history and examination, and medical records of clinically performed studies were reviewed, including chest radiographs, high-resolution computed tomography (HRCT) of the chest, and routine blood chemistry and hematologic tests. Some participants underwent pulmonary function testing, bronchoscopy with bronchoalveolar lavage (BAL) with examination of BAL cell cytology as reported (8) or surgical lung biopsy as part of their clinical care. When obtained, lung biopsies were processed and evaluated using standard methods. Disorders of surfactant production involving mutations of SFTPB, SFTPC, or ABCA3 were excluded by mutational analysis (Ambry Genetics or in the DNA Diagnostics Laboratory at the Johns Hopkins University, Baltimore, MD). All symptomatic patients underwent whole-lung lavage therapy (27) as indicated clinically. Case histories of the pediatric patients with PAP and details of the whole-lung lavage procedure are provided in the online supplement or were reported elsewhere (8, 28).

Serum GM-CSF autoantibody levels were measured by ELISA as reported (14, 16, 18). Commercial ELISA kits were used to measure SP-D (BioVendor, Candler, NC), GM-CSF (R&D Systems, Minneapolis, MN), monocyte chemotactic protein-1 (MCP-1; BD Biosciences, San Diego, CA), and macrophage colony–stimulating factor (M-CSF; R&D Systems, Minneapolis, MN) as directed by the manufacturers.

GM-CSF–stimulated levels of CD11b on blood leukocytes were evaluated as reported (14). Briefly, heparinized blood was incubated without or with GM-CSF (Leukine, Genzyme, Cambridge, MA; 10 ng/ml, 30 minutes, 37°C) and then CD11b levels were measured using flow cytometry.

GM-CSF–stimulated phosphorylation of STAT5 in blood leukocytes was evaluated as reported (8). Briefly, heparinized blood was incubated without or with GM-CSF (Leukine; 10 ng/ml, 15 minutes, 37°C). Cell lysates were evaluated by Western blotting using anti-STAT5 antibody (Santa Cruz Biotechnology, Santa Cruz, CA), anti–phospho-STAT5 (Millipore, Billerica, MA), or anti-actin (Santa Cruz Biotechnology) as primary antibodies.

Blood was collected by phlebotomy and GM-CSF receptor α and β chain levels on blood leukocytes were evaluated by flow cytometry using GM-CSF receptor α chain (BD Biosciences) or β chain (Santa Cruz Biotechnology) specific antibodies as reported (8).

Receptor-mediated GM-CSF binding was evaluated as reported (8). Briefly, peripheral blood mononuclear cells were isolated and seeded into culture dishes in medium containing recombinant human GM-CSF (Leukine, Berlex) at a concentration of 1 ng/ml. At various times, the GM-CSF concentration was measured in the culture medium by ELISA (R&D Systems) and expressed as a percentage of original GM-CSF concentration.

GM-CSF receptor α in blood leukocytes was evaluated by Western blotting using a GM-CSF receptor α chain–specific antibody (Santa Cruz Biotechnology) as reported (8).

Sequencing was performed in the Genetic Variation and Gene Discovery Core Facility at the Cincinnati Children's Hospital Medical Center. Briefly, genomic DNA and total RNA were obtained from blood leukocytes using commercial kits (QIAGEN, Valencia, CA) and polymerase chain reaction (PCR) was used to prepare DNA spanning each CSF2RA exon (using genomic DNA) or the entire coding sequence (using cDNA reverse transcribed from mRNA). Nucleotide sequences were compared with data reported in GenBank under accession NM_006140.3 (CSF2RA) or NM_000395.2 (CSF2RB).

Numeric data were evaluated for normality and variance using the Shapiro-Wilk and Levene median tests, respectively, and presented as the mean ± SE (parametric data) or median and interquartile range (nonparametric data). Statistical comparisons were made with Student t test, one-way analysis of variance, or Kruskal-Wallis rank-sum test as appropriate and P values less than or equal to 0.05 were considered to indicate statistical significance. Statistical analysis used SigmaPlot software (Systat, San Jose, CA). All experiments were repeated at least twice, with similar results.

Between May 2004 and November 2009, 187 individuals underwent testing for an increased serum GM-CSF autoantibody level to diagnose autoimmune PAP. Of these, 110 had a positive GM-CSF autoantibody test (>3 μg/ml), whereas five with a negative test were children (<18 years old) with PAP and no underlying disease associated with secondary PAP (7). Evaluation of the family members of these five children identified two more children with undiagnosed PAP and 16 healthy family members. One additional recently reported case (28) was included in the tabular analysis. The demographics and results of screening tests of the children with hereditary PAP are shown (Tables 1 and 2, respectively). The serum GM-CSF concentration in these children with GM-CSF autoantibody-negative PAP was higher than that of healthy family members and than that of unrelated healthy controls (Figure 1).

Figure 1. Serum granulocyte-macrophage colony–stimulating factor (GM-CSF) concentrations in children diagnosed with hereditary pulmonary alveolar proteinosis (hPAP), members of their immediate family who were healthy (family members), or healthy controls. The lower limit of quantification (LLOQ) of the assay was 7.8 pg/ml (dashed line). The serum levels of GM-CSF in children with hPAP (52 pg/ml [28–101 pg/ml]) were increased compared with healthy family members (0.0 pg/ml [0.0–3 pg/ml]) and unrelated healthy controls (0.0 pg/ml [0.0–1.9 pg/ml]) (n = 8, 11, 30, respectively; P < 0.001; Kruskal-Wallis analysis of variance on ranks with comparisons using Dunn's method).

Download Figure | Download Powerpoint

[More] [Minimize]

Eight children with PAP associated with recessive CSF2RA abnormalities (hereafter, patients) contributed to the clinical phenotype reported here (Table 1). Sixteen healthy family members of these patients were also evaluated, including individuals heterozygous for the molecular defects identified (Figure 2). All patients were full-term infants without any need for respiratory support at birth. Six (patients B–E, G, and H) developed progressive dyspnea, exercise intolerance, or tachypnea of insidious onset at various ages ranging from 1.5 to 9 years and two remained asymptomatic at age 5 (patient A) or 8 (patient F) years (Table 1). The mean ages at symptom onset, diagnosis of PAP, and report submission were 4.8 ± 1.6, 6.4 ± 1.2, and 8.6 ± 1.3, respectively. Disease severity at the time of diagnosis varied markedly, ranging from hypoxemic respiratory failure to asymptomatic. Poor weight gain and linear growth occurred in three (patients E–G) (Table 1). All patients had normal routine blood chemistry and hematologic indices. All symptomatic and one of two asymptomatic patients (patient F) had increased serum SP-D (Table 2). In addition to GM-CSF, the concentration of M-CSF and MCP-1 was increased in lung BAL fluid in all patients tested (Table 2). All symptomatic patients received whole-lung lavage therapy and experienced significant, although incomplete, clinical improvement (Table 1). Wide intersubject variability occurred in the number and frequency of whole-lung lavage therapy procedures needed.

Figure 2. Segregation of CSF2RA mutations in six families with hereditary pulmonary alveolar proteinosis (PAP). (A) Pedigree of patients A–C. (B) Pedigree of patient D. (C) Pedigree of patient E. (D) Pedigree of patients F and G. (E) Pedigree of patient H. Squares represent male family members and circles represent female members. Filled symbols represent persons with hereditary PAP. Allelic variants segregating within a family are shown above the pedigree. The status of both alleles is indicated under the symbols according to the following key referring to CSF2RA sequences: N/N = two normal alleles; changed nucleotide (or amino acid)/N = one mutated and one normal allele; changed nucleotide (or amino acid)/changed nucleotide (or amino acid) = two abnormal alleles; NA = not available for testing. The colloquial nomenclature for each mutation is shown here and the formal genetic names are given below (Table 3). The CSF2RA gene abnormalities caused by chromosomal deletions at Xp33.22 are indicated by the size of the deletion in megabases (mb): XpΔ1.6 = Xp33.22del1.6mb; XpΔ0.41 = Xp33.22del0.41 mb; Xq^Houston = Xq chromosome in patient H identified in Houston (28). Patient H had Turner syndrome (46 Xi(Xq). The CSF2RA XpΔ0.41 mutation originated on the paternal X chromosome and was transmitted to the normal length isochromosome X of patient H; her Xq CSF2RA allele was absent because of the deletion of Xp. The proband in each family is indicated (arrows). A question mark indicates the status of the second CSF2RA allele, if present, is unknown.

Download Figure | Download Powerpoint

[More] [Minimize]

HRCT of the chest was performed in all patients. Symptomatic individuals had characteristic diffuse bilateral ground-glass opacifications with sharply delimited margins bordered by fine lines forming polygonal shapes 3–10 mm in diameter (Figure 3). This “geographic” pattern reflected lobular and lobar boundaries and varied in intensity with disease severity. Patient A (asymptomatic at age 5) had limited abnormalities (Figure 3A) and patient F (asymptomatic at age 8) had scattered involvement of secondary lobules in each lobe (8). The chest radiographs of symptomatic patients showed bilateral, diffuse, patchy airspace disease (Figures 3B–E) similar to that reported for patient G (8).

Figure 3. High-resolution computed tomography scans of the chest in patients with hereditary pulmonary alveolar proteinosis obtained at the time of diagnosis. (A) Patient A at age 5 years and asymptomatic on room air. A small region of ground glass opacification is indicated (arrow). (B) Patient B at age 3 years and dyspneic on oxygen via nasal canula at 2 L per minute. (C) Patient C at age 4 years and dyspneic on oxygen via nasal canula at 2 L per minute. (D) Patient D at age 11 years and dyspneic on room air. (E) Patient E at age 11 years and dyspneic on 100% oxygen via facemask. All scans show findings at the level of the main carina.

Download Figure | Download Powerpoint

[More] [Minimize]

Pulmonary function testing was performed in patients D, F, and G at presentation and in patient E after an initial “rescue” whole-lung lavage treatment. All symptomatic patients evaluated had a restrictive ventilatory defect with proportionate impairment of the FVC, FEV₁, and TLC, and a disproportionately more severe reduction of the diffusing capacity for carbon monoxide (Dl_CO) (Table 1). The Dl_CO could not be performed in patient G, who had refused any breathholding maneuvers for several years before diagnosis. Patient F was asymptomatic and had normal spirometry but moderately reduced Dl_CO (Table 1).

All except patient F underwent bronchoscopy with BAL or whole-lung lavage with evaluation of BAL cytology (Table 1). Patients D, G, and H also underwent surgical lung biopsy. In all patients evaluated, the BAL fluid had a milky appearance with a substantial amount of sediment. In most, a dense accumulation of surfactant and cellular debris precluded adequate differential cytometry. In the asymptomatic patient who had bronchoscopy with BAL for diagnosis (patient A), the BAL cell differential included 79.6% macrophages, 4.2% neutrophils, 16% lymphocytes, and 0.2% eosinophils (n = 500 cells). Cytology also showed increased debris and numerous, enlarged foamy alveolar macrophages (Figure 4A) that stained positively with periodic acid–Schiff (PAS) reagent (Figure 4B), and oil red O (Figure 4C). Ultrastructural evaluation revealed that the foamy appearance of the alveolar macrophages was caused by the accumulation of numerous intracellular lamellar and lipid droplet inclusions (Figure 4D). Copious extracellular surfactant was also present (Figure 4D).

Figure 4. BAL cell cytology, alveolar macrophage ultrastructure, and lung histopathology in patients with hereditary pulmonary alveolar proteinosis. (A–D) Cytology and ultrastructure of bronchoalveolar lavage cells obtained from patient A at the time of diagnosis (5 yr). (A) Diff-Quick staining showing numerous enlarged, foamy alveolar macrophages (original magnification ×20, bar = 20 μm; inset original magnification ×100, bar = 2 μm). (B) Periodic acid–Schiff (PAS) staining showing glycoprotein accumulation in alveolar macrophages (original magnification ×20, bar = 20 μm; inset original magnification ×100, bar = 2 μm). (C) Oil red O staining and hematoxylin counterstaining showing accumulation of neutral lipids in alveolar macrophages (original magnification ×20, bar = 20 μm; inset original magnification ×100, bar = 2 μm). (D) Ultrastructural analysis demonstrating that the foamy appearance of alveolar macrophages is caused by the accumulation of lamellar body (open arrows) and lipid droplet (closed arrows) inclusions. The extracellular space had an abnormal abundance of surfactant (asterisk) (original magnification ×6,000, bar = 2 μm). (E–H) Histopathologic appearance of the lung biopsy obtained from patient D at the time of diagnosis (11 yr) after PAS staining (original magnification ×20). (E) Alveoli in some areas are completely filled with granular, PAS-positive material. (F) Region showing the sharp interface between alveoli filled with PAS-positive material and empty, normal alveoli. (G) Region showing well-preserved alveolar wall structure without abnormalities. (H) Region showing the accumulation of foamy alveolar macrophages (arrows) and small amounts of extracellular, intraalveolar PAS-positive material. Note the well-preserved alveolar walls.

Download Figure | Download Powerpoint

[More] [Minimize]

Microscopic examination of surgical lung biopsies showed that alveoli were filled with a fine granular, eosinophilic material staining strongly positive with PAS reagent (Figure 4E) and for SP-A, SP-B, and SP-D (8). The abnormal alveolar filling was patchy with completely filled alveoli adjacent to completely empty alveoli (Figure 4F), paralleling the geographic pattern of involvement visualized radiographically (Figure 3). Alveolar walls were well preserved and without abnormality (Figure 4G), similar to results for patients with autoimmune PAP (7), primates with PAP caused by injection of human PAP patient-derived GM-CSF autoantibodies (17), and mice deficient in GM-CSF (22) or its receptor (29). Less severely involved regions also showed intact, enlarged foamy alveolar macrophages in alveoli with no or only minor amounts of alveolar filling with the granular material (Figure 4H). Scattered regions of increased lymphocytes were occasionally seen (8). These results contrast sharply with lung biopsies of patients with surfactant production disorders caused by mutations in SFTPB, SFTPC, or ABCA3, which routinely show gross parenchymal distortion, epithelial cell hyperplasia, and variable accumulation of dysfunctional surfactant (30).

All symptomatic patients underwent whole-lung lavage therapy as described in the online supplement, which resulted in clinical and radiographic improvement. Radiographic clearing was evident by comparison of routine chest radiographs before (Figure 5A) and after (Figure 5B) whole-lung lavage therapy. The degree of radiographic improvement was more clearly defined on HRCT scans of the chest before (Figure 5C) and after (Figure 5D) whole-lung lavage therapy. Although all patients improved with this therapy, the degree of overall improvement varied between washes, among patients, and between lavage of the right and left lungs (the right was generally more difficult). Patients received a mean (± SEM) number of whole single-lung lavages of 8 ± 2.3 with 10.5 ± 1.4 L per lung each over a period of 1.8 ± 0.4 years. Importantly, the number of lavages required did not correlate with the age at diagnosis or duration of disease (R² = −0.303, −0.273, respectively; patients A–G; P = 0.491; Spearman rank order correlation).

Figure 5. Radiographic appearance of the lungs in patients with hereditary pulmonary alveolar proteinosis before and after whole-lung lavage. Posteroanterior chest radiographs of patient E are shown (A) before and (B) 1 week after the second of two bilateral whole-lung lavage procedures (13 L saline each) separated by 3 months. The oxyhemoglobin saturation was 90% on 100% oxygen before the first and 97% on room air after the second procedure. Note the marked clearing of the infiltrate in both lungs and improved visualization of the cardiac border after lavage. High-resolution computed tomography scans of the chest of patient C are shown (C) before and (D) 8 weeks after a right whole single-lung lavage with 5.35 L and 12 weeks after left whole single-lung lavage with 3.78 L of saline. Note the marked clearing of the infiltrates in both lungs after lavage.

Download Figure | Download Powerpoint

[More] [Minimize]

Having established that GM-CSF signaling dysfunction was not caused by neutralizing GM-CSF autoantibodies, we evaluated the patients for GM-CSF receptor abnormalities based on the hypothesis that this might cause GM-CSF accumulation as it does in mice with GM-CSF receptor defects (8). GM-CSF receptor function was evaluated by measuring leukocyte levels of either phosphorylated STAT5 (8) or cell-surface CD11b (14) in whole blood incubated without and with GM-CSF. Patients A–G had no GM-CSF–stimulated increase in phosphorylated STAT5 (Figure 6A; see Figure E1 in the online supplement) (8) and none of the patients had an increase in GM-CSF–stimulated CD11b levels (Table 2). In contrast, controls and healthy family members all had readily detectable GM-CSF–stimulated increases in phosphorylated STAT5 (Figure 6A; see Figure E1) or CD11b (not shown) (8, 28) in blood leukocytes.

Figure 6. Analysis of granulocyte-macrophage colony–stimulating factor (GM-CSF) receptors in patients with hereditary pulmonary alveolar proteinosis. Fresh heparinized blood was obtained from patient B, her unaffected parents, and an unrelated healthy control and blood leukocytes were evaluated for GM-CSF signaling (A) or GM-CSF receptor α chains on the cell surface (B) or in cell lysates (C). (A) Blood was incubated without (-) or with (+) GM-CSF (10 ng/ml, 15 minutes) followed by Western blotting to detect phosphorylated STAT5 (pSTAT5), total STAT5 (STAT5), or actin (to ensure equal loading of cell lysates). As a positive control for STAT5 phosphorylation in patient leukocytes, cells were incubated with IL-2, which resulted in readily detectable STAT5 phosphorylation as expected (42) (not shown). (B) Blood was immunostained and flow cytometry was used to detect GM-CSF receptor α protein on the cell surface of blood leukocytes from patient B, her father and mother, or an unrelated healthy control (C). In each histogram, the y axis scale represents 0–40 cells per histogram cell and the x axis scale represents 10–1,000 fluorescence intensity units. Note the absence of GM-CSF receptor α for patient B and its presence for family members and control. All patients, family members, and controls had readily detected GM-CSF receptor β on blood leukocytes (not shown). (C) Blood was used to prepare leukocyte lysates, which were fractionated on polyacrylamide gradient gels and subjected to Western analysis to detect and determine the molecular mass of GM-CSF receptor α. Western analysis of parallel samples for actin ensured equal loading of cell lysates. The band corresponding to GM-CSF receptor α glycoprotein is broad because of glycosylation at multiple sites (43). Note the absence of GM-CSF receptor α in patient B but not her parents or control.

Download Figure | Download Powerpoint

[More] [Minimize]

To determine if the absence of either GM-CSF receptor α or β chains was the basis for the absence of GM-CSF signaling, blood leukocytes were evaluated by flow cytometry. GM-CSF receptor β was present on blood leukocytes from all patients and the mean fluorescence intensity was similar to that of healthy family members and healthy controls (Table 2). We confirmed the function of the β chain in patients by stimulation of blood leukocytes with IL-3, which resulted in STAT5 phosphorylation as expected (31).

In contrast, GM-CSF receptor α was not detectable in patients A–E or patient H, but was present on leukocytes from all healthy family members and controls evaluated (Figure 6B; see Figure E2). GM-CSF receptor α was also evaluated by Western blotting of leukocyte lysates, which confirmed its absence in patients B–E and presence in family members and healthy controls (Figure 6C; see Figure E3). The GM-CSF receptor α was detectable on leukocytes from patients F and G but had a reduced molecular mass on polyacrylamide gels because of the consequences of disruption of a single glycosylation site (8). Clearance of GM-CSF by blood leukocytes was absent in patients B and C and markedly reduced in patients F and G (Table 2), consistent with a defect in GM-CSF binding.

Thus, the absence (patients A–E and H) or structural abnormality (patients F and G) of the GM-CSF receptor α protein provided an explanation for the impaired GM-CSF signaling in these patients: both a severe reduction and the complete absence of GM-CSF signaling are associated with the development of PAP.

Nucleotide sequence analysis of genomic DNA and mRNA from blood leukocytes of each patient, their family members, and healthy controls was done to determine the molecular nature of abnormal CSF2RA gene expression. The sequencing strategy and primers used are shown (Figure 7A and Table E1, respectively). For clarity, results from the propositus in each family are described together with their respective family members. Patient B was homozygous for a single nucleotide substitution (C to T) at position 649 (i.e., c.649C > T) in CSF2RA exon 8 (Figure 7B). This point mutation introduced a premature stop codon at position 217 (i.e., Arg217X). Analysis of mRNA confirmed the genomic DNA abnormality in patient B (not shown). Both healthy parents and two healthy siblings (AIII1, AIII2, AIV1, and AIV2, respectively; Figure 2) were heterozygous for this point mutation and the other healthy sibling (AIV5) had only a normal sequence. Patient A, the brother of patient B (Figure 2), was homozygous for this mutation in genomic DNA and mRNA (not shown, but identical to Figure 7B). Patient C, a first cousin of patients A and B, was homozygous for this mutation (not shown, but similar to Figure 7B), whereas of both her healthy parents and two healthy siblings (AIII3, AIII4, AIV7, and AIV8, respectively; Figure 2) were heterozygous for it and a normal CSF2RA sequence (not shown). The remainder of the CSF2RA coding sequence of mRNA from patients A–C was normal (not shown).

Figure 7. Analysis of CSF2RA mutations in patients with hereditary pulmonary alveolar proteinosis. (A) Schematic of CSF2RA and the polymerase chain reaction (PCR) amplification and strategy used for sequencing. The locations of Alu sequence high copy number genomic repeat sequence motifs are indicated (hatched boxes). (B) Partial nucleotide sequence of CSF2RA exon 8 in genomic DNA from patient B and a healthy control. Numbering is according to the CSF2RA reference sequence in Genbank (accession No. NM_006140.3). (C) Partial nucleotide sequence CSF2RA exon 10 in mRNA from patient D and a healthy control. (D) PCR amplification of the CSF2RA coding sequence from mRNA from patient E, her father (F), mother (M), brother (B), or a healthy individual (C). (E) PCR amplification of CSF2RA exons 1–13 in genomic DNA from patient E and family members (indicated as in D). (F) PCR amplification of genomic DNA from patient E and family members (indicated as in D) in the region surrounding the Alu family repeats flanking CSF2RA exon 7 as shown in the schematic in A. Note the small band in the patient, small and large bands in each parent, and only the large band in the brother. (G) Partial nucleotide sequence of CSF2RA intron 6 (black arrows) and exon 7 (red arrow) in genomic DNA from patient E and a healthy control. (H) Partial nucleotide sequence of CSF2RA exons 6–10 in mRNA from patient E and a healthy control showing a single mRNA transcript sequence in the healthy control and two overlapping transcript sequences beginning at Lys158 in the patient. (I) Proposed model for the generation of the CSF2RA mutation present in patient E (homozygous) and her parents (heterozygous). See text for details. The locations of the two Alu family repeats flanking Exon 7 are indicated (hatched regions). These regions are 90% homologous at the DNA level, and contain a 10-nucleotide central region of 100% homology (black region within the hatched region).

Download Figure | Download Powerpoint

[More] [Minimize]

Analysis of leukocyte mRNA from patient D revealed a single nucleotide sequence containing a two-nucleotide (GC) duplication immediately after nucleotide position 921 (i.e., c.920_921dupGC) in exon 10, resulting in a reading frame shift at position 308 and introduction of a premature stop codon at position 332 in exon 11 (Figure 7C). Analysis of genomic DNA confirmed the mRNA abnormality in patient D (not shown). The remainder of the CSF2RA coding sequence of mRNA from patient D was normal (not shown). Analysis of genomic DNA and mRNA sequences showed that her mother (BI2; Figure 2) and grandmother (not shown) did not harbor this mutation. The father was unavailable for evaluation, and CSF2RA allelic heterozygosity analysis was not performed.

In patient E, reverse transcription and PCR amplification of mRNA showed that the CSF2RA coding sequence was approximately 200 nucleotides shorter than that of her brother (or a healthy control), and that both parents (CII1, CI1, and CI2, respectively; Figure 2) had PCR products similar in size to the patient and to control (Figure 7D). This suggested a CSF2RA deletion mutation may be segregating in this family. All CSF2RA exons were detected in genomic DNA from patient E and her family members (and a healthy control) except for exon 7, which was detected in all except the patient (Figure 7E). Targeted PCR amplification showed patient E was homozygous for a 1.7-kb deletion encompassing exon 7 and that both parents were heterozygous carriers of this mutation, whereas her brother was homozygous normal (Figure 7F). Analysis of genomic DNA sequence showed the deletion included 1,794 nucleotides (c.474–1007_646+614del1,794) encompassing exon 7 (Figure 7G). Sequence analysis revealed two tandem Alu family repetitive DNA sequences flanking exon 7 in the parents and brother of patient E and a healthy control (Figure 7A). These Alu repeats were 90% homologous with each other and contained a central 10-nucleotide region of 100% homology (Figure 7G, overlapping region of red and black arrows) among the family members and control. Patient E had a deletion extending from within the central homologous region in the upstream 5′ flanking Alu repeat to within this same central sequence in the downstream 3′ flanking Alu repeat. Analysis of mRNA sequence revealed that leukocytes from patient E expressed two CSF2RA mRNA transcripts encoding distinct protein abnormalities (Figure 7H). Transcript 1 comprised a precise deletion of exon 7, substitution of threonine for arginine at codon 159, followed by a frame shift introducing a premature stop codon 56 codons downstream within exon 10 (i.e., p.Arg159ThrfsX56). Transcript 2 comprised the deletion of exon 7 and the first 19 nucleotides of exon 8, resulting in substitution of asparagine for lysine at position 158 followed by an in-frame deletion of the next 65 codons (i.e., p.Lys158_Ser222delinsAsn). Comparison of genomic DNA and mRNA sequence revealed these were alternatively spliced transcripts (Figure 7I): transcript 1 used the natural donor site of exon 6 and the natural splice acceptor site for exon 8 and transcript 2 used this same donor and a cryptic splice acceptor site within exon 8, 19 nucleotides downstream from the 5′ end of exon 8. Together, these findings strongly suggest that homologous recombination between these two Alu repeats in an ancestral CSF2RA gene (as observed in γ^o-β^o-thalassemia [32, 33]) is the mechanism underlying the CSF2RA exon 7 deletion mutation in this family (Figure 7I).

These results identify that a wide range of genetic CSF2RA abnormalities and their corresponding protein abnormalities (Figure 8) are associated with development of hereditary PAP. These included CSF2RA missense, nonsense, duplication, deletion, and frameshift mutations; aberrant alternative splicing; gene deletions; and chromosomal rearrangements with partial gene deletion (Table 3).

Figure 8. CSF2RA mutations and granulocyte-macrophage colony–stimulating factor (GM-CSF) receptor α abnormalities associated with the development of hereditary pulmonary alveolar proteinosis. (Top) Schematic of the normal GM-CSF receptor α protein showing regions corresponding to the signal peptide, extracellular GM-CSF binding domain, transmembrane spanning domain, and short intracellular, cytoplasmic domain, along with the 11 N-glycosylation sites (43). The asterisk indicates the glycosylation site disrupted by the G196R mutation. (Middle) Schematic of the CSF2RA mRNA showing exons 1–13. The start codon (ATG) and stop codon (TGA) are indicated. (Bottom) Schematic representations of the effects of each CSF2RA mutation (indicated at left) on the GM-CSF receptor α protein structure are shown. The colloquial nomenclature is used for each CSF2RA abnormality. In each, stippled bars represent the signal peptide, black bars represent the extracellular domain, gray bars represent the transmembrane domain, white bars represent the cytoplasmic domain, hatched bars represent regions in which the reading frame is shifted, and dotted bars represent regions that are missing in the predicted protein product. Bars indicate the size of a peptide 50 amino acids in length (50 aa) or a nucleotide sequence 100 nucleotides in length (100 nt).

Download Figure | Download Powerpoint

[More] [Minimize]

Patients with hereditary PAP presented with clinical manifestations because of the effects of surfactant accumulation on gas exchange analogous to the presentation of autoimmune PAP patients (7) except at a younger age (51 [n = 223] vs. 6.4 ± 1.2 [n = 6] years, respectively). Although not designed to study prevalence, we identified 110 cases of autoimmune PAP and seven cases of hereditary PAP among 187 sera screened for autoimmune PAP suggesting that hereditary PAP represents a relatively small proportion of total cases of PAP caused by loss of GM-CSF signaling (∼6% in our series). The disease occurred in both sexes, white and black individuals, and on three different continents. The radiologic appearance was similar among symptomatic patients and identical to that of autoimmune PAP patients (7, 19). Of diagnostic importance was the observation that the radiograph in hereditary PAP appeared worse than expected based on the clinical findings. Our data suggest that radiographic changes in hereditary PAP progress slowly, likely preceding the development of symptoms by months or years. Although pulmonary function tests (especially Dl_CO) can be useful in assessing disease severity, they are difficult to perform in children less than 6 years old. Routine laboratory tests were of little help. However, several novel biomarkers were diagnostically useful. Serum GM-CSF autoantibody levels, which are increased in all patients with autoimmune PAP (7, 15), were low in all patients with hereditary PAP. In contrast, GM-CSF levels were elevated in all patients with hereditary PAP and low in patients with autoimmune PAP (16). A substantial delay in diagnosis (∼2 yr) and misdiagnosis as asthma (before radiography) or pneumonia (after radiography) underscores the importance of a high degree of clinical suspicion in making a timely and accurate diagnosis of hereditary PAP. Our results support the value of performing HRCT of the chest, serum GM-CSF autoantibody, GM-CSF, and SP-D testing in pediatric patients with PAP at the time of diagnostic assessment. If bronchoscopy with BAL is performed, elevated levels of GM-CSF, MCP-1, or M-CSF may be useful. Evaluation of GM-CSF–stimulated increases in leukocyte CD11b levels or STAT5 phosphorylation by flow cytometry are convenient screening tests to evaluate for GM-CSF signaling dysfunction. If this strategy is used, patients with hereditary PAP can be effectively identified without the need for surgical lung biopsy. Lastly, HRCT may be a useful clinical outcome measure in the evaluation of novel therapies for PAP because it directly measures the accumulation of the pathogenic material (surfactant) in the lungs. This conclusion is supported by a recent report using quantitative CT densitometry to measure the response to GM-CSF therapy in autoimmune PAP (40).

Patients with hereditary PAP responded well to whole-lung lavage, similar to results for patients with autoimmune PAP (7, 19, 24) and in contrast to patients with disorders of surfactant production who do not respond well. Of technical importance, the narrower airway in children precludes the use of a standard Carlen double-lumen tube and special materials and procedures were needed (see the online supplement). No formal criteria were developed to measure the therapy responses in our study and these should be developed and evaluated in future studies. Based on our prior use of quantitative CT densitometry to monitor outcomes (40), we believe a limited scan chest HRCT protocol may be useful. Serial serum SP-D measurement may also be of use. Bone marrow transplantation was attempted in one patient who died shortly after the procedure of uncontrolled respiratory infection (28). This approach has been considered for some of our current patients; however, suitable donors have been difficult to find. We successfully demonstrated the feasibility of gene therapy for GM-CSF receptor defects in mice (41) and in preliminary in vitro human studies (8). Further studies are required to evaluate these potential therapies.