Abstract
Pulmonary alveolar proteinosis (PAP) is a rare disease characterized by the accumulation of phospholipids and surfactant proteins in the lung. The central role for granulocyte-macrophage colony-stimulating factor (GM-CSF) in surfactant homeostasis has been established in mice lacking the GM-CSF gene, which results in murine pulmonary alveolar proteinosis. No GM-CSF gene defect has been defined in adult patients with idiopathic PAP. Previous studies indicated that the human disease differs from the murine model by the presence of circulating, neutralizing autoantibodies against GM-CSF. Therefore, the final common pathway between the GM-CSF knockout and human PAP appears to be the deficiency of functionally active GM-CSF. In the present study, all patients with idiopathic PAP were found to have systemic and localized antibodies against GM-CSF. Anti–GM-CSF titers were a specific and sensitive marker for PAP. In addition, we present data showing that the absence of active GM-CSF is associated with enhanced levels of macrophage colony-stimulating factor, monocyte chemoattractant protein-1, and interleukin-8. These studies confirm and strengthen previous studies and support the concept that adult idiopathic PAP is an autoimmune disease defined by the presence of anti–GM-CSF. Further, using anti–GM-CSF as an indicator of pulmonary alveolar proteinosis may avoid the use of more invasive means of evaluating patients with pulmonary disease characterized by alveolar infiltrates.
Adult pulmonary alveolar proteinosis (PAP) is a rare disease of unknown etiology characterized by the accumulation of phospholipids and surfactant proteins in the alveoli. It has been proposed that the development of PAP is due to the inability of the alveolar macrophages and type II epithelial cells to clear the excess surfactant (1, 2). The diagnosis of PAP usually requires an open lung biopsy, and the standard therapy for PAP is the physical removal of the accumulated surfactant by whole-lung lavage (3).
The central role for granulocyte-macrophage colony-stimulating factor (GM-CSF) in surfactant homeostasis has been established in murine models by several lines of evidence. First, mice lacking the GM-CSF gene present with impaired surfactant clearance leading to murine alveolar proteinosis, which resembles human PAP (4). Second, the development of alveolar proteinosis in these mice can be corrected by lung-specific delivery of the GM-CSF gene, (5) aerosolization of GM-CSF (6), or bone marrow transplantation for hematopoeitic reconstitution (7). Studies by Kitamura and coworkers, and other studies, indicate that the human disease differs from the murine model in that patients have circulating, neutralizing autoantibodies against GM-CSF (8–11) while retaining an intact GM-CSF gene. Whether the autoimmune response is specific for GM-CSF is unknown. In addition, a subset of PAP patients improve with GM-CSF therapy (12, 13), which supports the hypothesis that the absence of GM-CSF either by gene-knockout or antibody-mediated neutralization results in the development of PAP.
We hypothesize that PAP is an autoimmune disease defined by the specific induction of anti–GM-CSF. The purpose of the current study is to evaluate the presence of circulating anti–GM-CSF as a specific and sensitive marker for PAP and determine whether the detection of systemic GM-CSF antibodies in patients presenting with pulmonary infiltrates is diagnostic for PAP.
This study was approved by the Institutional Review Board, and informed consent was obtained from all subjects. The diagnosis of PAP was established in all patients by open lung biopsy, or by characteristic findings on imaging studies and bronchoscopy with bronchoalveolar lavage (BAL) and biopsy (n = 40). The characteristics of patients with PAP were as follows: age, 38 ± 11 yr (mean ± SD); 26 male/14 female; 26 diagnosed by open lung biopsy, 14 diagnosed by transbronchial biopsy; and 30 had whole lung lavages ranging from 1–33 times. None of the patients with PAP were on medications at the time samples were obtained. The PAP in these patients was felt to be idiopathic in all cases based on exclusion of known secondary causes of PAP. Healthy volunteers (n = 21) were identified by local advertising and through the employee pool. Disease controls (n = 23) included patients with high anti-nuclear antibody (ANA) titer with no pulmonary disease (n = 5), beryllium lung disease (granuloma positive, n = 4), asthma (n = 4), pneumonia (n = 3), pulmonary metastatic cancer (n = 5), and patients with bone marrow disease on long-term GM-CSF therapy (n = 2). The patients with ANA titer, pneumonia, pulmonary metastatic cancer, and bone marrow disease were on multiple medications, whereas our patients with beryllium and asthma were not on any medications.
BAL was performed as previously described (10). Fluid was separated from cells, aliquoted, and stored at –80°C until assayed. PAP BAL was evaluated undiluted and compared with healthy and disease control BAL. Cytospins were prepared from BAL cells and stained with Wright-Giemsa (Sigma, St. Louis, MO).
Serum samples were obtained from all patients with PAP and control subjects as previously described (10, 12). Briefly, peripheral blood was collected in serum separator tubes, centrifuged, aliquoted, and stored at –80°C until tested. PAP sera was evaluated over several dilutions and compared with healthy and disease control serum samples.
Enzyme-linked immunosorbent assay (ELISA) plates were coated with 20 μg/ml recombinant human GM-CSF (Immunex, Seattle, WA) in Na(HCO3) overnight at 4°C. The plates were washed then blocked overnight at 4°C with tris-buffered saline (TBS) containing 5% nonfat dry milk followed by serial dilutions of patient and control serum. Because anti–GM-CSF autoantibodies are frequently detected in pharmacologic preparations of human IgG (14, 15), we used 200 μg/ml IVIg (Immuno, Rochester, MI) as a control for naturally occurring anti–GM-CSF. In addition, an anti–GM-CSF positive PAP serum was added as a positive control. The IVIg and patient sample were used to standardize and control for variability. In the assay, the samples and standards were incubated for 90 min, followed by several washes and the addition of anti-human Ig (Fab') labeled with horseradish peroxidase (HRP; Jackson ImmunoResearch, SanDiego, CA) for 60 min. The HRP-labeled antibody was removed and the plates were washed, followed by the addition of chromogenic substrate (Pierce, Rockford, IL). The results were evaluated at 450 nm. End titers were established by determining the point at which the optical density (OD) of antibody dilution was equivalent to those observed for the healthy control subjects. The assay for macrophage colony-stimulating factor (M-CSF) was identical, using recombinant M-CSF (R&D Systems, Minneapolis, MN).
BAL fluid from patients with PAP, disease control subjects, and healthy volunteers was assayed by ELISA for M-CSF, GM-CSF, interleukin (IL)-3, IL-8, monocyte chemoattractant protein-1 (MCP-1) (R&D Systems), granulocyte colony-stimulating factor (G-CSF; Endogen, Woburn, MA), and GM-CSF receptor (GM-CSFR; Immunotech, Marseille, France). The sensitivity of the assays ranged from 31–1,000 pg/ml for M-CSF, 31–1,000 for IL-3, 39–2,500 for G-CSF, 20–1,000 for MCP-1, 26–1,000 for IL-8, 15–250 for GM-CSF, and from 20–640 pM for GM-CSFR. Samples below the sensitivity were reported as the lowest value and samples above the sensitivity were diluted to obtain a value within the linear assay range. Recovery experiments were done using GM-CSF ELISA Kits (R&D, Endogen). Briefly, serum from patients with PAP, disease control subjects, and healthy control subjects was spiked with three different concentrations of GM-CSF based upon the linear part of the assay's standard curve. Samples were run in duplicate and percent (%) recovery was determined as compared with baseline GM-CSF levels in the serum. The coefficient of variation was < 10%. All assays were performed in duplicate.
Statistical analysis was performed by a nonparametric one-way ANOVA test using GraphPad Prism version 3.00 for Windows (GraphPad Software, San Diego, CA). Significance was defined as P ⩽ 0.05. Interpreting the serum titer for laboratory diagnosis was done using the formulas defined by Motulsky and coworkers (16). The sensitivity is the fraction of all those with the disease who get a positive test result. The specificity is the fraction of those without the disease which get a positive test result. Predictive values in the population as a whole were not estimated due to the unknown incidence of PAP in the general population.
All our adult patients with idiopathic PAP (n = 40) had detectable anti–GM-CSF, whereas little or no anti–GM-CSF was in the serum of healthy or disease control subjects (Figure 1)
Figure 1. All patients with idiopathic PAP have systemic anti–GM-CSF. Serum from patients with PAP (PAP, n = 40), healthy control subjects (HC, n = 21), and disease control subjects (DC, n = 23) were evaluated for anti–GM-CSF auto-antibodies. PAP serum contained significant levels of anti–GM-CSF (P < 0.0001 as compared with disease and healthy control subjects).
[More] [Minimize]The next focus was to investigate the local presence of anti–GM-CSF by evaluating BAL. BAL from 20 of 20 patients with PAP had end-titers of anti–GM-CSF at ⩾ 1:100 (Figure 2)
Figure 2. All patients with idiopathic PAP have localized levels of anti–GM-CSF. BAL fluid from patients with PAP (PAP, n = 20), healthy control subjects (HC, n = 6), and disease control subjects (DC, n = 6) were evaluated for the presence of anti–GM-CSF. BAL fluid from patients with PAP had significantly more anti–GM-CSF (P < 0.0001 as compared with disease and healthy control subjects).
[More] [Minimize]To establish specificity of the anti–GM-CSF response in PAP, we looked for the presence of other colony-stimulating factors: M-CSF, IL-3, and G-CSF in BAL (Figure 3)
Figure 3. BAL fluid contains elevated levels of M-CSF. BAL fluid was obtained from patients with PAP (PAP, n = 28), healthy control subjects (HC, n = 5), and disease control subjects (DC, n = 5) and evaluated for the presence of M-CSF, IL-3, and G-CSF. Significant levels of M-CSF were found in PAP BAL (P = 0.0375). BAL IL-3 and G-CSF were not detected.
[More] [Minimize]BAL fluid obtained from patients with PAP had significantly more detectable MCP-1 (Figure 4A)
Figure 4. BAL fluid contains elevated levels of MCP-1 and IL-8. BAL fluid was obtained from patients with PAP (PAP, n = 28), healthy control subjects (HC, n = 5), and disease control subjects (DC, n = 5) and evaluated for the presence of MCP-1, IL-8, and MIP-1. Significant levels of MCP-1 (P < 0.001) and IL-8 (P < 0.001) were found in PAP BAL. There was no detectable MIP-1.
[More] [Minimize]
Figure 5. PAP BAL differentials are heterogeneous with lymphocytes, neutrophils, and eosinophils. BAL cells were obtained from patients with PAP (PAP, n = 17) and healthy control subjects (HC, n = 7). Cells were stained and differentials were obtained. PAP BAL cells were heterogenous with significant numbers of alveolar macrophages, lymphocyte, and neutrophils (P < 0.001) compared with > 95% alveolar macrophages in healthy control subjects.
[More] [Minimize]There was no difference between patients with PAP, healthy control subjects, and disease control subjects in systemic GM-CSFR levels (Figure 6)
Figure 6. GM-CSFR concentration in PAP, disease, and healthy control serum is similar. Serum from patients with PAP (PAP, n = 8), healthy control subjects (HC, n = 4), and disease control subjects (DC, n = 4) were analyzed using a commercially available GM-CSFR kit. There was no difference in concentration of GM-CSFR between PAP serum samples and control subjects (P = 0.6608).
[More] [Minimize]The presence of GM-CSF was determined in sera and BAL of patients with PAP and control subjects. The GM-CSF levels were detected in only 5/28 PAP BAL samples, but not in the BAL from the control subjects (Figure 7)
Figure 7. PAP BAL fluid contains GM-CSF. BAL from patients with PAP (PAP, n = 28) was evaluated for GM-CSF. Low levels of GM-CSF were found in five of the BAL samples and no GM-CSF was found in the healthy (HC, n = 5) or disease (DC, n = 5) control samples.
[More] [Minimize]Our data shows that all adult patients with idiopathic PAP have systemic and localized antibodies against GM-CSF. The systemic end-titers were ⩾ 1:400 and localized end-titers were ⩾ 1:100 in PAP, whereas healthy and disease control subjects had little or no detectable end-titers. We also investigated whether abnormalities in other hematopoeitic growth factors might be present in patients with PAP. Interestingly, patients with PAP have extremely elevated levels of M-CSF in the lung; however, even with these elevated levels of M-CSF, anti–M-CSF was not detected in the serum or the BAL. M-CSF is a differentiation/activation cytokine acting to stimulate macrophage survival and differentiation (17). The elevation of M-CSF has also been reported in the GM-CSF knockout mouse (18), suggesting that it may be a compensatory mechanism due to a deficiency in biologically active GM-CSF. Other hematopoeitic growth factors (IL-3 and G-CSF) were not elevated, suggesting a specific relationship between GM-CSF and M-CSF, potentially due to their many shared and complimentary functions (19). The absence of detectable anti–M-CSF suggests that the anti–GM-CSF response is specific in PAP and not part of a generalized autoimmune process. Serum anti–GM-CSF end-titers of ⩾ 1:400 were diagnostic for PAP in our patient population.
In addition to the elevated presence of M-CSF, we found the lungs of patients with PAP to contain significant levels of the chemokines MCP-1 and IL-8. Iyonaga and colleagues also showed elevated BAL MCP-1 in patients with PAP (20). The presence of MCP-1 and IL-8 may potentially contribute to the increased numbers of lymphocytes and neutrophils observed in the PAP BAL, which occurred in the absence of lung infection. The GM-CSF knockout mouse also has elevated levels of BAL MCP-1 (21), suggesting a primary role for this cytokine in lung homeostasis. The presence of chemokines in the lungs of PAP patients may create an environment favorable for cellular recruitment as observed in Figure 5, which may be either primary or secondary to the anti–GM-CSF autoimmunity.
Previous reports have indicated that many patients with PAP have circulating anti–GM-CSF, which may neutralize GM-CSF activity (8–11). Kitamura and coworkers first showed that the GM-CSF antibody was present in 11 (9) and 24 (8) patients with PAP, and suggested that neutralization of GM-CSF may be involved in the pathogenesis of the disease. Most recently, this group, in collaboration with Seymour and colleagues (22), also evaluated a group of Australian patients with PAP and found all had circulating anti–GM-CSF. These investigators further characterized the auto-antibody as neutralizing and polyclonal, reacting to anti-human IgG1 and IgG2 but not IgG3 or IgG4 (8). In the initial studies, a competitive GM-CSF ELISA was used to determine the presence of anti–GM-CSF. These studies were followed by the development of anti–GM-CSF blotting and latex agglutination test. The anti–GM-CSF assay utilized in the present study demonstrates a sensitivity of 100% and an estimated conservative specificity of 91%. The sensitivity and specificity of the assay were defined based upon anti–GM-CSF end-titer of healthy control serum (< 1:10), disease control serum (< 1:100), or IVIg (< 1:100). Therefore, our data indicate that elevation of systemic anti–GM-CSF (⩾ 1:400) is diagnostic of PAP, and may be a simple noninvasive test for the initial diagnosis of PAP.
Unlike the murine models, the auto-antibody in human PAP suggests that it is not the absence of GM-CSF but the potential deactivation of the cytokine that is important. In fact, in a subset of patients with PAP a low level of GM-CSF was detected by ELISA, but these levels were apparently not related to any clinical parameters such as BAL differential or anti–GM-CSF titer. Deactivation of GM-CSF could occur by antibody neutralization or soluble receptor blocking of available GM-CSF. PAP serum completely blocked the detection of spiked GM-CSF, suggesting that the anti–GM-CSF complexes with the GM-CSF. Whether the ELISA recognizes receptor bound or free GM-CSF in PAP serum remains to be determined. The fact that the soluble GM-CSFR levels are not different between disease and healthy control serum samples, and that it is not present in BAL, suggests that GM-CSFR does not play a major role in blocking the GM-CSF bioactivity either systemically or in the lung. As reported by Kitamura and colleagues and Thomassen and coworkers, complete neutralization of GM-CSF by PAP sera but not by healthy control serum samples was diagnostic of PAP (9, 10), even with the similar levels of GM-CSFR.
Because all patients with PAP have circulating levels of neutralizing anti–GM-CSF (8, 10), we propose that deficient GM-CSF in PAP is the result of direct neutralization and deactivation of the cytokine by GM-CSF auto-antibodies. GM-CSF is a key factor for promoting the differentiation and proliferation of macrophages (23), an essential cell involved in surfactant clearance. Dysfunction of alveolar macrophages due to neutralizing the biologic effects of GM-CSF is a plausible mechanism in the development of PAP. To our knowledge, this is the first direct association between an antibody response to an essential cytokine and a specific disease. We believe that not only does this data contribute to this hypothesis, but also that the presence of anti–GM-CSF is a specific and sensitive marker for PAP. The detection of systemic anti–GM-CSF offers a simple, noninvasive diagnostic test for PAP without the risks associated with the current diagnostic standard of open lung biopsy. Studies are underway to determine whether the antibody titer correlates with disease activity, response to therapy, or is predictive of prognosis. Further, our data support the previous observations that GM-CSF plays a critical and essential role in healthy lung homeostasis.
The authors would like to thank Dr. Barbara Barna for comments and review of the manuscript. This work was funded by NIH grant HL67676.
| 1. | Milleron, B. J., U. Costabel, H. Teschler, R. Ziesche, J. L. Cadranel, H. Matthys, and G. Akoun. 1991. Bronchoalveolar lavage cell data in alveolar proteinosis. Am. Rev. Respir. Dis. 144:1330–1332. |
| 2. | Mazzone, P., M. J. Thomassen, and M. Kavuru. 2001. Our new understanding of pulmonary alveolar proteinosis: what an internist needs to know. Clev. Clin. J. Med. 68:977–992. |
| 3. | Shah, P. L., D. Hansell, P. R. Lawson, K. B. M. Reid, and C. Morgan. 2000. Pulmonary alveolar proteinosis: clinical aspects and current concepts on pathogenesis. Thorax 55:67–77. |
| 4. | Stanley, E., G. J. Lieschke, D. Grail, D. Metcalf, G. Hodgson, J. A. M. Gall, D. W. Maher, J. Cebon, V. Sinickas, and A. R. Dunn. 1994. Granulocyte/macrophage colony-stimulating factor-deficient mice show no major perturbation of hematopoiesis but develop a characteristic pulmonary pathology. Proc. Natl. Acad. Sci. USA 91:5592–5596. |
| 5. | Zsengeller, Z. K., J. A. Reed, C. J. Bachurski, A. M. LeVine, S. Forry-Schaudies, R. Hirsch, and J. A. Whitsett. 1998. Adenovirus-mediated granulocyte-macrophage colony-stimulating factor improves lung pathology of pulmonary alveolar proteinosis in granulocyte-macrophage colony-stimulating factor-deficient mice. Hum. Gene Ther. 9:2101–2109. |
| 6. | Reed, J. A., M. Ikegami, E. R. Cianciolo, W. Lu, P. S. Cho, W. Hull, A. H. Jobe, and J. A. Whitsett. 1999. Aerosolized GM-CSF ameliorates pulmonary alveolar proteinosis in GM-CSF-deficient mice. Am. J. Physiol. 276:L556–L563. |
| 7. | Nishinakamura, R., R. Wiler, U. Dirksen, Y. Morikawa, K. Arai, A. Miyajima, S. Burdach, and R. Murray. 1996. The pulmonary alveolar proteinosis in granulocyte macrophage colony-stimulating factor/interleukins 3/5 bc receptor-deficient mice is reversed by bone marrow transplantation. J. Exp. Med. 183:2657–2662. |
| 8. | Kitamura, T., K. Uchida, N. Tanaka, T. Tsuchiya, J. Watanabe, Y. Yamada, K. Hanaoka, J. F. Seymour, O. D. Schoch, I. Doyle, Y. Inoue, M. Sakatani, S. Kudoh, A. Azuma, T. Nukiwa, T. Tomita, M. Katagiri, A. Fujita, A. Kurashima, S. Kanegasaki, and K. Nakata. 2000. Serological diagnosis of idiopathic pulmonary alveolar proteinosis. Am. J. Respir. Crit. Care Med. 162:658–662. |
| 9. | Kitamura, T., N. Tanaka, J. Watanabe, K. Uchida, S. Kanegasaki, Y. Yamada, and K. Nakata. 1999. Idiopathic pulmonary alveolar proteinosis as an autoimmune disease with neutralizing antibody against granulocyte/macrophage colony-stimulating factor. J. Exp. Med. 190:875–880. |
| 10. | Thomassen, M. J., T. Yi, B. Raychaudhuri, A. Malur, and M. S. Kavuru. 2000. Pulmonary alveolar proteinosis is a disease of decreased availability of GM-CSF rather than an intrinsic cellular defect. Clin. Immunol. 95:85–92. |
| 11. | Tanaka, N., J. Watanabe, T. Kitamura, Y. Yamada, S. Kanegasaki, and K. Nakata. 1999. Lungs of patients with idiopathic pulmonary alveolar proteinosis express a factor which neutralizes granulocyte-macrophage colony-stimulating factor. FEBS Lett. 442:246–250. |
| 12. | Kavuru, M. S., E. J. Sullivan, R. Piccin, M. J. Thomassen, and J. K. Stoller. 2000. Exogenous granulocyte-macrophage colony-stimulating factor administration for pulmonary alveolar proteinosis. Am. J. Respir. Crit. Care Med. 161:1143–1148. |
| 13. | Seymour, J. F., A. R. Dunn, J. M. Vincent, J. J. Presneill, and M. C. Pain. 1996. Efficacy of granulocyte-macrophage colony-stimulating factor in acquired alveolar proteinosis. N. Engl. J. Med. 335:1924–1925. (Letter) |
| 14. | Svenson, M., M. B. Hansen, C. Ross, M. Diamant, K. Rieneck, H. Nielson, and K. Bendtzen. 1998. Antibody to granulocyte-macrophage colony-stimulating factor is a dominant anti-cytokine activity in human IgG preparations. Blood 91:2054–2061. |
| 15. | Ullenhag, G., C. Bird, P. Ragnhammar, J. E. Frodin, K. Strigard, A. Olsterborg, R. Thorpe, H. Mellstedt, and M. Wadhwa. 2001. Incidence of GM-CSF antibodies in cancer patients receiving GM-CSF for immunostimulation. Clin. Immunol. 99:65–74. |
| 16. | Motulsky, H. 1995. Interpreting Lab Tests: Introduction to Bayesian Thinking. In Intuitive Biostatistics. Oxford University Press, New York. 129–148. |
| 17. | Deng, P., Y. L. Wang, V. L. Shahbazian, and P. K. Pattengale. 2000. Biological characterization of uncleavable plasma membrane-anchored human macrophage colony-stimulating factor. Biochem. Biophys. Res. Commun. 276:304–311. |
| 18. | Shibata, Y., Y. P. Berclaz, Z. C. Chroneos, M. Yoshida, J. A. Whitsett, and B. C. Trapnell. 2001. GM-CSF regulates alveolar macrophage differentiation and innate immunity in the lung through PU.1. Immunity 15:1–20. |
| 19. | Roth, P., and E. R. Stanley. 1992. The biology of CSF-1 and its receptor. Curr. Top. Microbiol. Immunol. 181:141–167. |
| 20. | Iyonaga, K., M. Suga, T. Yamamoto, H. Ichiyasu, M. Miyakawa, and M. Ando. 1999. Elevated bronchoalveolar concentrations of MCP-1 in patients with pulmonary alveolar proteinosis. Eur. Respir. J. 14:383–389. |
| 21. | Paine, R., S. B. Morris, H. Jin, S. E. Wilcoxen, S. M. Phare, B. B. Moore, M. J. Coffey, and G. B. Toews. 2001. Impaired functional activity of alveolar macrophages from GM-CSF-deficient mice. Am. J. Physiol. Lung Cell. Mol. Physiol. 281:L1210–L1218. |
| 22. | Seymour, J. F., J. J. Presneill, O. D. Schoch, G. H. Downie, P. E. Moore, I. R. Doyle, J. M. Vincent, K. Nakata, T. Kitamura, D. Langton, M. C. Pain, and A. R. Dunn. 2001. Therapeutic efficacy of granulocyte-macrophage colony-stimulating factor in patients with idiopathic acquired alveolar proteinosis. Am. J. Respir. Crit. Care Med. 163:523–531. |
| 23. | Dranoff, G., A. D. Crawford, M. Sadelain, B. Ream, A. Rashid, R. T. Bronson, G. R. Dickersin, C. J. Bachurski, E. L. Mark, J. A. Whitsett, and R. C. Mulligan. 1994. Involvement of granulocyte-macrophage colony-stimulating factor in pulmonary homeostasis. Science 264:713–716. |
