|Home | About | Journals | Submit | Contact Us | Français|
IPF is a devastating disease with few therapeutic options. The precise aetiology of IPF remains elusive. However, our understanding of the pathologic processes involved in the initiation and progression of this disease is improving. Data on the mechanisms underlying IPF have been generated from epidemiologic investigations as well as cellular and molecular studies of human tissues. Although no perfect animal model of human IPF exists, pre-clinical animal studies have helped define pathways which are likely important in human disease. Epithelial injury, fibroblast activation and repetitive cycles of injury and abnormal repair are almost certainly key events. Factors which have been associated with initiation and/or progression of IPF include viral infections, abnormal cytokine, chemokine and growth factor production, oxidant stress, autoimmunity, inhalational of toxicants and gastro-oesophageal reflux disease. Furthermore, recent evidence identifies a role for a variety of genetic and epigenetic abnormalities ranging from mutations in surfactant protein C to abnormalities in telomere length and telomerase activity. The challenge remains to identify additional inciting agents and key dysregulated pathways that lead to disease progression so that we can develop targeted therapies to treat or prevent this serious disease.
The pathogenesis of IPF is complex. The initiation of pulmonary fibrosis likely stems from a variety of factors that all result in a common pathologic outcome. Numerous epidemiologic risks have been postulated as inciting factors; however, no unifying etiologic agent or process has been identified. In an effort to more clearly define the pathogenesis of IPF, research has expanded from the clinical realm to also include evaluation of molecular and genetic pathways. The heterogeneity of IPF has become a striking difficulty in elucidating the pathways involved in its development. In fact, more recent reviews of this disease have proposed that IPF represents a syndrome rather than a specific disease entity. It is possible that many cases of IPF represent unrecognized rheumatologic, environmental or occupational toxic exposures, or genetic diseases. This is highlighted by the finding that the underlying pathology of IPF, usual interstitial pneumonitis (UIP), may be precipitated by a variety of diseases such as hypersensitivity pneumonitis, scleroderma or rheumatoid arthritis. In addition to the clinical and molecular variability of IPF, many of the original studies investigating IPF likely included patients with other idiopathic diseases such as non-specific interstitial pneumonitis (NSIP) that may have very different pathogenesis, and therefore respond quite differently to treatment. Epidemiological, genetic, molecular and clinical studies are often inherently limited or flawed because of this apparent heterogeneity, and therefore are extremely difficult to compare. Regardless of the initiating cause, the mechanisms that lead to UIP/IPF are poorly understood. These mechanisms may or may not be unique to the underlying disease precipitating the fibrosis. In this review article, we will explore the proposed mechanisms for the initiation of IPF and discuss the clinical and molecular indicators that may predict progression of this disease (Fig. 1).
Viral infections have been widely postulated as initiators of fibrosis. Viruses have been intrinsically difficult to study because of the variable sensitivity of respiratory sample cultures. The advent of PCR and molecular detection methods has dramatically increased the yield of viral detection. However, this increase in sensitivity has raised concerns that the new methods of viral detection are overly sensitive. Because of these concerns, the causal relationship of viruses and fibrosis has become a matter of contest. While many viruses have been detected in the sputum, bronchoalveolar lavage fluid (BALF) and lung tissue of patients with IPF, the question lingers whether the IPF makes patients more susceptible to viral infection or whether the viral infection initiated the fibrosis.
Several viruses have been implicated in the pathogenesis of IPF including Epstein–Barr Virus (EBV), Human Herpes Viruses 7 & 8, cytomegalovirus (CMV), Hepatitis C virus (HCV), Herpes simplex virus, parvovirus B19 and TT virus. EBV has been the most widely investigated for its contribution to pulmonary fibrosis. EBV is very common, typically infecting more than 95% of people over their life span. The prevalence of detectable EBV in patients with IPF varies by geographical populations. However, EBV isolation from respiratory samples is not specific to IPF. EBV has been isolated at high percentages in patients with other fibrotic lung diseases such as scleroderma.1 A study from Scotland showed that WZhet, a rearranged EBV DNA typically associated with the active replication and lytic phase of EBV infection, was found at higher frequencies in patients with IPF. This particular EBV genome rearrangement was detected in 61% of lung tissue specimens and 59% of buffy coats in their cohort of IPF patients.2 There was no correlation between WZhet and immunosuppressive therapy. While EBV PCR positivity has been shown to be higher in sporadic IPF lung tissue specimens as mentioned above when compared with familial IPF, there was no difference between rates of PCR positive lung tissue, suggesting that viral infection may simply be more common due to the underlying structural lung disease.
The role of CMV in the pathogenesis of IPF has also been widely investigated. In a review of 102 lung transplant recipients, only five were CMV positive. All of these patients had IPF.3 In a population of patients with IPF in Poland, a high prevalence of CMV (75%) was detectable in the BALF, blood leukocytes and serum. While this was no different from the rate of positive healthy controls, the CMV DNA copy number was significantly higher in the serum of patients with IPF.4 Furthermore, serum complement fixation titres for CMV were elevated in patients with IPF and collagen vascular disease when compared with sarcoidosis, emphysema and healthy controls.5 Comparison of viral PCR between sporadic and familial IPF in the southern United States showed significantly higher prevalence of CMV in familial IPF.6
Hepatitis C virus has also been implicated as a contributor in the pathogenesis of IPF. In Japan, there were significant differences between IPF (28.8%) and the general population (3.66%).7 This was corroborated by an Italian study showing HCV antibodies were detectable in 13% of patients with IPF compared with 0.3% of blood donor controls and 6.1% in a non-interstitial lung disease control group.8 However, in a study from the United Kingdom, the prevalence of HCV infection in patients with IPF was not higher than in the general population.9 None of these studies showed a clear association between IPF and liver disease. Furthermore, a study comparing HCV-infected patients with EBV-infected patients concluded that there was a higher rate of IPF in patients infected with HCV, 0.3% at 10 years and 0.9% at 20 years, compared to no patients with EBV infection. Factors increasing the likelihood of developing IPF included smoking, age greater than 55 and presence of cirrhosis.10
The TT virus is a recently identified single stranded DNA virus, originally discovered in a patient with post transfusion hepatitis of unknown aetiology. It has recently been isolated from 12 of 33 (36.4%) Japanese patients with IPF. Several different genome types of the TT virus were present in four of these patients.11 The actual role of the TT virus remains unclear.
While all of these viruses have all been associated with IPF, they are also found at higher frequencies in other interstitial lung diseases and in patients who have been treated with corticosteroids, suggesting a possible link between treatment of these disorders and infection or reactivation of the viruses.12 Comparison of studies evaluating the role of viruses in IPF remains difficult. It is not always possible to classify differences in sample handling, viral identification techniques have various sensitivitities, and the rates of viral positivity vary on the basis of type of biological sample as well as regional prevalence of various viruses. Due to these formidable challenges, there has never been a direct causal relationship established between viruses and IPF.
Gene targets from familial IPF have been investigated to gain insight into the role genetics may play in the development of fibrosis. A genome scan with subsequent fine mapping of six families with familial IPF revealed loci of interest, and a previously uncharacterized ELMOD2 gene.13 ELMOD2 is expressed in the lung and has significantly decreased mRNA expression in the lungs of IPF patients compared with healthy controls. Based on these results, ELMOD2 is a candidate gene for susceptibility in familial IPF that requires further investigation.
Surfactant Protein C (SP-C) has also been shown to be lacking or decreased in BALF of some patients with familial IPF.14 SP-C has been shown to be important to postnatal lung function and therefore is implicated for the development of lung dysfunction such as IPF.15 One heterozygous mutation of SP-C in exon 5 was noted in a large family with 11 affected members further indicating its importance in familial IPF.16 Two SP-C mutations have been identified that cause misfolding of proteins and subsequent type-II epithelial cell damage, which may provide a possible mechanism for disease.14
Genetic alterations have also been noted in sporadic IPF. In a case–control study two MMP-1 gene promoter polymorphisms were investigated: 2G and T/G SNP polymorphisms.17 Both polymorphisms are associated with activator protein 1 (AP-1), which has previously been shown to be associated with a risk of IPF. Both the 2G and T/G SNP polymorphisms were more frequent in IPF patients than control after controlling for smoking status. This suggests the importance of genetic polymorphisms of the MMP-1 gene promoter as well as the importance of the environmental pressure of smoking on genetic predispositions in determining risk for developing IPF. Erythrocyte complement receptor 1 (CR1) is another factor of interest that has been shown to be abnormal in other fibrotic and inflammatory diseases such as sarcoidosis. Similar to other complement components, it is important in the uptake of immune complexes into phagocytes. It was hypothesized that CR1 would also be important to developing IPF.18 Comparing three CR1 polymorphic sites in IPF patients and controls, significant linkage disequilibrium was found for one genotype, which is consistent with the proposed role of CR1.19
Epigenetic processes change DNA or transcribed proteins in a heritable but reversible manner. Thy-1 is a membrane protein expressed on multiple cells including fibroblasts and has been shown to suppress fibroblast differentiation into myofibroblasts, which are associated with fibrotic processes.20 Using methylation-specific PCR to identify epigenetically downregulated Thy-1 in human lung tissues, methylated Thy-1 promoters were found in fibroblastic foci in IPF. Additionally, treatment with DNA methyltransferase restored Thy-1 expression in Thy-1(-) fibroblasts presumably by removing the epigenetically imposed suppression.21 Together, these data indicate epigenetic Thy-1 suppression may contribute to pulmonary fibrosis.
Telomere shortening represents a genetic phenomenon that is associated with ageing via alteration of regular tissue renewal. Telomerase components have been hypothesized to play a role in IPF given the increased incidence of IPF with increasing age and the hypothesis that IPF develops as a consequence of alterations in tissue regeneration, such as re-epitheliazation. This may be particularly true in patients with a family history of IPF.22,23 Leukocyte and alveolar epithelial cell telomere length in IPF patients without a family history of IPF was shorter compared with healthy controls.24 These studies strongly suggest that telomere shortening is a marker for and may contribute to pulmonary fibrosis.
Derangement of pulmonary surfactant and alveolar collapse are seen in IPF.25 BALF SP-A is decreased and Serum SP-A and SP-D are increased in patients with IPF and may have roles in progression.25,26 SP-A and SP-B genetic polymorphism variants were associated with IPF indicating each may serve as markers of patients at risk for IPF.27 This association was not seen for SP-C or SP-D.
The cytokine and chemokine milieu in the serum and lung tissue may be essential to the development of pulmonary fibrosis.28,29 Numerous cytokines and chemokines have been identified as potential contributors to the initiation of fibrosis, as they are notably elevated in BALF and/or whole lung tissue of patients with IPF. Other cytokines have been identified through the examination of resident cells in the lung such as macrophages, fibroblasts, neutrophils, and alveolar epithelial cells and comparing differences between cells from patients with IPF and controls. Although a clear pathway has not been identified, cytokines and chemokines likely direct a complicated concert of cellular and extracellular matrix interactions that lead to fibrosis and subsequently drive its progression.
Several growth factors have been identified as potential contributors to IPF. These include transforming growth factor-β (TGF-β), insulin-like growth factor (IGF-1), platelet derived growth factor (PDGF) and connective tissue growth factor (CTGF). In general, growth factors act to promote cell differentiation and proliferation. Their effects tend to be cell-specific and therefore they may exert different influences on different cell types. Because of these complicated interactions, the precise role of each of these growth factors is not entirely clear.
Transforming growth factor-β has been implicated as a central factor in the development of pulmonary fibrosis. TGF-β has a wide range of biological effects, in particular, the promotion of wound repair via an increase in extracellular matrix deposition, recruitment of inflammatory cells and the promotion of fibroblast differentiation.28,30–32 This is highlighted by the fact that in vivo overexpression of active TGF-β in rat lungs via an adenovirus vector induces fibrosis characterized by the presence of myofibroblast cells and extracellular matrix deposition.31 Additionally, bleomycin and silica induced lung fibrosis have been ameliorated in animal models of fibrosis by a variety of anti-TGF-β strategies such as decorin, byglycan, peroxisome proliferator receptor gamma ligands (PPARγ) and soluble TGF-β receptors.33–38 The cellular responses to TGF-β are variable, but in general TGF-β drives both epithelial cells and fibroblasts towards a fibrotic phenotype. Heightened responses to TGF-β occur in fibroblasts from patients with IPF compared with controls, suggesting that TGF-β alone is not the only reason for fibrosis progression.39 Epithelial cells are driven to differentiate into a mesenchymal cell phenotype.40–42
In human IPF, TGF-β gene expression is elevated in fibroblasts from the lungs of patients with IPF compared with controls,43 and levels of TGF-β are elevated in BALF fluid from patients with IPF. However, this is not specific to IPF, as BALF levels of TGF-β are also elevated in stage IV sarcoidosis and systemic sclerosis.44 TGF-β is localized to alveolar epithelial cells, alveolar macrophages, the bronchial epithelium, and is associated with the extracellular matrix.45 In particular, the epithelial cells lining honeycomb cysts stained intensely for TGF-β. Although a similar pattern of TGF-β overexpression was also seen in NSIP, the intensity of TGF-β immunohistochemical staining was much stronger in UIP.46,47 Together with the animal data, these findings strongly suggest TGF-β as a key cytokine that induces fibrosis.
Insulin-like growth factor is another mediator that has been implicated in the development of IPF. It has many physiologic functions including cell migration and differentiation and it is elevated in the lungs of IPF patients.30,48 IGF-1, which is produced by macrophages, lymphocytes and possibly epithelial cells, may contribute to re-epithelialization by acting as an antiapoptotic factor.30 IGF-1 has also been shown to stimulate extracellular matrix production by stimulating fibroblast collagen synthesis in vitro.49,50 The responses of fibroblasts obtained from patients with IPF appear to have different responses to IGF-1 than fibroblasts obtained from healthy controls. Compared with normal fibroblast strains, fibroblasts from the lungs of IPF patients exhibit heightened synthetic activity, particularly enhanced fibronectin production. This is compared with a predominantly proliferative response seen in the healthy control fibroblasts.51 IGF-1’s interactions are often mediated through multiple binding proteins. IGF binding proteins (IGFBP)-3 and −5 are increased in vivo in IPF lungs and in vitro in corresponding fibroblasts, suggesting that they also play a significant role in the development of fibrosis.52 IGFBP-3 and −5 may also increase the activity of IGF-1 and subsequent fibrotic and re-epithelialization responses mentioned above.53
Platelet derived growth factor is another growth factor that is elevated in the BALF and lung tissue from patients with IPF that parallels expression with IGF-1 localization. In the early stages of IPF, PDGF and IGF-1 proteins were localized to alveolar macrophages, mononuclear phagocytes, fibroblasts, alveolar Type II cells, vascular endothelial cells and vascular smooth-muscle cells. In the later stages of IPF, the localization of PDGF and IGF-1 proteins was similar to healthy controls, predominantly localized in alveolar macrophages;48,54 however, the number of PDGF positive interstitial macrophages was significantly increased.55 In vitro proliferative responses to PDGF suggest no differences between UIP fibroblasts versus controls.56 Similar to IGF-1, PDGF enhanced synthesis of fibronectin in IPF fibroblasts compared with controls.51 Another possible role of these cytokines involves the development of angiogenesis and pulmonary hypertension. This theory is related to the localization of IGF-1 and PDGF in vascular endothelial cells and smooth muscle cells.57 Despite these observations, the roles of PDGF and IGF-1 in fibrosis remain unclear. Their contribution to the complex cytokine interactions requires further investigation.
Connective tissue growth factor is a pro-fibrotic cytokine that increases expression of extracellular matrix deposition and induces fibroblast proliferation.58 CTGF has been localized to type II alveolar cells and interstitial fibroblasts in lung tissue from IPF.59 CTGF is upregulated by TGF-β60 and therefore acts downstream of TGF-β to direct the fibrotic process.61 While CTGF is likely an important mediator of the fibrotic process, increases in CTGF are probably not sufficient to drive fibrosis on its own. CTGF overexpression in animal models resulted in mild reversible fibrosis.62 Further investigation into the interaction of TGF-β and CTGF is necessary to better define its role in initiation of pulmonary fibrosis.
IL-1β is a pro-inflammatory cytokine important to wound repair. This cytokine has been implicated in the pathogenesis of IPF, particularly because of its in vivo effects in animal models of fibrosis. Intratracheal IL-1β adenoviral gene transfer in rat lungs resulted in increased pro-inflammatory cytokines IL-6 and Tumour Necrosis Factor-α (TNF-α) as well as increases in the pro-fibrotic growth factors PDGF and TGF-β.63 Furthermore, in vivo data show that fibroblasts treated with IL-1β differentiate to myofibroblasts and increase extracellular matrix (ECM) deposition. The role of IL-1β in human disease is less clear. There are varying results on whether IL-1β is increased in the lungs of patients with IPF. Rather, it may be the IL-1 receptor antagonist (IL-1ra) that may be more biologically important. Higher levels of IL-1ra without compensatory increases in IL-1β have been found in BALF from patients with IPF.64 Current smokers with IPF had lower levels of IL-1ra, suggesting a possible link between epidemiologic data showing smokers with IPF have better survival rates.
Tumour Necrosis Factor-α is a key pro-inflammatory cytokine that is upregulated in pulmonary fibrosis.28 Alveolar macrophages produce TNF-α in response to LPS treatment,65 and levels of TNF-α are increased in mice following radiation exposure66 and in moth eaten mutant mice prone to fibrosis.67 Furthermore, TNF-α expression along with TGF-β expression, was induced in mice treated with bleomycin.68 Overexpression of TNF-α in mice and rats leads to fibrosing alveolitis69,70 and TNF-α receptor knockout mice are protected from asbestos-induced fibrosis.71 Other data using TNF-α receptor knockout mice show that TNF-α expression but notTGF-β expression was increased following bleomycin exposure. These mice also failed to exhibit pulmonary toxicity or fibrosis related to bleomycin.68 These data suggest an important interplay between TNF-α and TGF-β. These models suggest a central role of TNF-α to the development of pulmonary fibrosis pathway.
IL-6 is another cytokine that is induced by TGF-β.72 Overexpression of IL-6 in mice demonstrated a prominent peribronchiolar mononuclear cell response, as well as a mild increase in fibrosis.73 Variable responses to IL-6 are exhibited by fibroblasts from IPF compared with normal controls in vitro. A proliferative response to IL-6 is seen in IPF fibroblasts while there is an anti-proliferative effect seen in normal fibroblasts.74 IL-11 shares similar homology with IL-6 and is also elevated in the lungs of IPF localized to alveolar epithelial cells, airway smooth muscle cells and infiltrating macrophages. Similar to IL-6, IL-11 expression is also induced by TGF-β.75 In contrast to IL-6, IL-11 induced proliferation in both IPF and normal fibroblasts.74 Overexpression of IL-11 in mice also shows prominent peribronchiolar mononuclear inflammation. However, in contrast to IL-6, IL-11 also induced marked deposition of collagen I and collagen III along with increased numbers of myofibroblasts.76,77 Further investigation into these cytokines is necessary to better define their roles in human lung disease.
IL-8, a cytokine that mediates neutrophil chemotaxis, is another inflammatory cytokine that is elevated in BALF and serum from patients with IPF.78,79 Specifically, IL-8 production localizes to alveolar macrophages and may contribute to the patchy neutrophil influx seen in pulmonary fibrosis.80 Given the paucity of neutrophilic inflammation, it is understandable that IL-8, amongst other inflammatory cytokines, by themselves is insufficient to produce fibrosis. Overexpression of IL-8 and Regulated upon Activation, Normal T-cell Expressed, and Secreted (RANTES) do not induce fibrosis in rodent lung.62
In addition to the neutrophilic cytokine profile outlined above, a T helper cell type 2 (Th2) inflammatory profile that includes IL-4 and IL-5, is also present in IPF.81 Immunohistochemistry on infiltrating cells within the interstitium of IPF patients shows mononuclear cells stained for IL-4 and IL-5.82 Macrophage colony-stimulating factor (M-CSF) and CC Chemokine Ligand 2 (CCL2)/Monocyte chemoattractant protein-1 (MCP-1) are also both elevated in BALF of IPF patients compared with control.83,84 This finding is corroborated by the observation that M-CSF and CCL2 knockout mice have less pulmonary fibrosis following bleomycin administration.84 The importance of the Th2 cytokine profile and mononuclear phagocyte infiltration in pulmonary fibrosis remains unclear, but these data suggest that both lymphocytes and monocytes may have an important role in the disease process.
Prostaglandin E2 (PGE2) is also hypothesized to play a role in IPF. In vitro data show that PGE2 suppresses fibroblast proliferation, differentiation and collagen synthesis.85–88 Fibroblasts cultured from IPF lung tissue showed less PGE2 production in response to IL-1β and LPS89 and TGF-β90 compared with healthy controls. These findings are corroborated by increased levels of fibrosis in cyclooxygenase 2 (COX2) deficient mice.91 COX2 overexpression in mice increases PGE2 expression in the lung and subsequently decreases fibroblast proliferation.92 Despite this convincing evidence for diminished PGE2 levels contributing to the development of IPF, there are data that show elevated PGE2 levels in BALF of patients with IPF.93 Further investigation into the role of PGE2 in IPF is warranted.
An immunologic aetiology of IPF has been hypothesized for decades; however, identification of a unifying antigen or antibody remains elusive. Nevertheless, there are significant proportions of patients with IPF that display various degrees of autoantibody positivity. Anti-nucleolar antibodies, while typically associated with rheumatologic disease, are present in approximately 25% of patients with IPF.94 In a subset of Anti-nucleolar antibodies (ANA) positive patients, Fischer et al. described the presence of antibodies to Anti Th/To, antibodies that are found in 4–13% of patients with scleroderma, in 13 of 25 (52%) IPF patients. Four of the 13 met criteria for limited cutaneous scleroderma (CREST variant scleroderma) and nine met criteria for SSC sine scleroderma.94 Even when ANA serology is negative, other autoantibodies may be detectable. In a small cohort of IPF patients, two out of eight patients had persistently positive antibodies against alveolar type II cells, reinforcing the role of epithelial cells in the pathogenesis of IPF.95
Antibodies against fibroblasts and/or extracellular matrix have also been implicated in the pathogenesis of IPF. These include antibodies to vimentin, cytokeratin 8, cytokeratin 18, type III procollagen, type III collagen and type I collagen. Serum anti-vimentin levels were elevated in patients with IPF compared with healthy controls; however, they were also elevated in patients with NSIP, suggesting a lack of specificity for IPF.96 Serum anti-cytokeratin 8 and 18 levels and cytokeratin: anti-cytokeratin 8 and 18 immune complex levels were significantly increased in patients with IPF. They were also elevated in BALF and serum from patients experiencing an acute exacerbation of IPF compared with healthy controls. The specificity of these antibodies for IPF is unclear.97–99 Serum collagen antibodies were detected in 13/16 (81%) of IPF patients.100 Similar antibodies to type III procollagen have been detected at higher levels in BALF from patients with IPF when compared with healthy controls and sarcoidosis,101 but again, the specificity for IPF is not known.
Topoisomerases, enzymes that modify intranuclear DNA, have been implicated in the pathogenesis of numerous genetic diseases. They currently serve as targets for multiple drug classes including antibiotics and anti-tumour agents. Serum autoantibodies to DNA topoisomerase II α have been demonstrated in patients with IPF.102 The reactivity of recombinant proteins with the autoantibody in serum correlated with disease duration but not severity.103 This supports not only the role of autoantibodies to DNA topoisomerase II α in pulmonary fibrosis but also implicates genetic alterations as critical to the development of IPF. Additional evidence supporting the importance of epigenetic regulation of pulmonary fibrosis has been elaborated by examining the role of telomeres. Telomere activity, measured as telomerase reverse transcriptase (TERT) activity, is upregulated during fibroblast differentiation in the bleomycin models of fibrosis.104,105 Fibroblasts isolated from TERT mice exposed to bleomycin showed less alpha smooth muscle actin expression and less pulmonary fibrosis. Bone marrow transplant of wild type marrow restored telomerase activity and levels of fibrosis returned to that of controls.106
Vascular immunology has also been implicated in the pathogenesis of IPF. In a small cohort of patients with IPF with no evidence of collagen vascular disease, anti-phospholipid antibodies were present in all 18 patients tested. These included anti-phosphatidylethanolamine, anti-phosphatidylcholine and anti-phosphatidylserine antibodies. In addition, anti-Ro and/or anti-Ribonuclear peptide (RNP) antibodies were seen in four of these patients.107 There are no data yet that correlate vascular antibodies with the development of pulmonary hypertension. Several other novel antibodies have been detected in IPF. Antibodies to IL-1α,108 p53,109 alanyl-tRNA synthetase110 and the Thompson-Friedrich antigen111 have also been identified in IPF. The significance of these antibodies remains to be determined.
While the prevalence of these highlighted antibodies raises the possibility of a causal relationship, this has yet to be determined. The autoantibodies may merely be a reactive process to a dysregulated biological system. It is entirely possible that the underlying pathology of UIP compromising IPF is more accurately a heterogeneous group of diseases, some of which may be autoimmune in nature.
Fibroblasts and myofibroblasts are the resident structural cells of the lung. They typically originate from mesenchymal cells; however, newer data suggest that some of these cells may in fact come directly from the bone marrow112,113 or from circulating pleuripotent cells called fibrocytes.114 The myofibroblast is a key effector cell in IPF, and is thought to originate from fibroblasts, possibly through exposure to TGF-β.115–117 It is also theorized that myofibroblasts may also originate from alveolar epithelial cells,118 a process referred to as epithelial-mesenchymal transition. New data also suggest that myofibroblasts may originate directly from the bone marrow or from circulating fibrocytes.119 Myofibroblasts of bone marrow origin have been found in the lung after the development of radiation fibrosis,120 fibrosis secondary to scleroderma,121 rheumatoid arthritis and mixed connective tissue disease,122 as well as in models of asthma airway remodelling.123,124 Therefore, bone marrow derived fibroblasts or fibrocytes are not likely to be specific to IPF, but rather constitute a systemic process of wound repair.
Fibrocytes represent <1% of the circulating population, and are distinguished through expression of ECM markers, particular Cluster of differentiation (CD) markers, MHC class II markers and chemokine receptors.125–129 The role of fibrocytes in IPF is unclear; however, fibrocytes have been localized to fibroblastic foci from patients with IPF,130 suggesting they play an integral role in the pathologic hallmark of this disease. Fibrocytes recruited to areas of lung injury likely contribute to pro-fibrotic cytokine or chemokine generation, particularly TGF-β.127
Animal models of fibrosis have shown similar fibrocyte recruitment to areas of lung injury as illustrated by fluoroscein isothiocyanate (FITC) lung injury in mice.131 In a severe combined immunodeficiency bleomycin mouse model of fibrosis, human fibrocytes also trafficked to the lung, the primary area of injury.132 Senescence-accelerated prone mice compared with age-matched control senescence-accelerated resistant mice had significantly higher levels of circulating and recruited fibrocytes after bleomycin injury, suggesting there are age-related alterations in fibrocyte populations.133 This correlates well with the increase in incidence of IPF with advancing age.
Due to the broad spectrum of disease that results from inhalation of toxic dusts or fumes, there has been an intense focus on the possibility that IPF represents an environmental or occupational lung disease. The association of IPF with tobacco smoke has been documented in many epidemiological studies and is consistently identified as a notable risk factor in multiple geographical populations.134–139 Epidemiological data also suggest increased odds ratios (OR) for developing IPF after exposure to birds, livestock, wood dust, metal dust, and stone cutting or sand. Other occupations associated with an increased OR include hair dressing, mining and agriculture/farming.134,138,140,141 There are inherent difficulties related to establishing causative occupational or environmental exposures from these epidemiological studies. These include the inability to correct for variable degrees of exposures, a dose–response relationship and recall bias. While there appears to be an association between occupational exposures and IPF, a causal relationship has not been determined.
The importance of oxidant injury in the development of IPF was recognized by the demonstration of an increase in oxidized BALF proteins142 and increased oxidative metabolism of alveolar macrophages resulting in reactive oxygen intermediates.143,144 Further investigation of alveolar macrophages in IPF showed increased production of pro-inflammatory pro-fibrotic leukotrienes B4 and C4 when compared with controls,145 suggesting that the macrophages in IPF exist in an activated state and may be and integral part of oxidant injury. Still other studies have demonstrated abnormal antioxidant injury localized to fibroblastic foci, epithelial cells and neutrophils. Extracellular superoxide, the major antioxidant enzyme found in the extracellular matrix, was found to be absent in areas of fibrosis.146 Inducible nitric oxide synthetase and the by-product of peroxynitrite oxidation, nitrotyrosine, were both elevated in epithelial cells, neutrophils and alveolar macrophages in lung tissue from patients with IPF versus controls.147 Furthermore, evaluation of BALF, induced sputum and serum have shown lower levels of glutathione in patients with IPF.148–150 Other in vitro studies showed that glutathione inhibits fibroblast proliferation in vitro and that TGF-β, a pro-fibrotic cytokine, directly decreased glutathione levels in fibroblasts, thereby providing a possible mechanism for the progressive injury, proliferation and cell differentiation found within the fibroblastic foci.151
The incidence of IPF increases with age,152–154 but it is unclear whether lead time bias contributes to this observation, as early fibrotic lung disease may be overlooked particularly in younger asymptomatic patients. Many diseases exhibit an age-associated increase in their incidence; however, there are relatively few clinical diseases that epidemiological surveys have identified as being more prevalent in patients with IPF. Gastro-oesophageal reflux diseases (GERD), diabetes and vascular disease are a few diseases that have shown an association with IPF. Several studies have highlighted the fact that there is a high incidence of GERD in patients with IPF, ranging from 66% to 94%. The importance of this revelation was further highlighted in one study by the recognition that most patients with IPF had asymptomatic GERD,155–157 and that 12 out of the 19 patients still had oesophageal reflux as measured by pH probe despite treatment with standard doses of proton pump inhibitors.156
Patients with diabetes had an elevated OR (4.06) for the development of IPF in a cohort of Japanese patients. There were no other identifiable differences in the clinical characteristics of the patients with IPF that could account for the development of diabetes.137 Vascular disease, but not diabetes, was associated with IPF in a cohort of subjects in the United Kingdom. The incidence of acute coronary syndrome (OR 1.53), angina (OR 1.84) and deep venous thrombosis (OR 1.98) were all increased compared with community matched controls in the 12 months preceding the diagnosis of IPF. During the follow-up period of 2–4 years, there was an adjusted increase in the risk of acute coronary syndrome (OR 3.14) and Deep venous thrombosis (DVT) (OR 3.39).158 It is not clear, however, whether IPF is contributing to the increased rates of vascular disease, or whether there is a common basis for these diseases.
Hubbard et al. evaluated the role of commonly prescribed medications on the development of IPF. IPF was associated with the use of antidepressants with an OR of 1.79.159 A follow-up large epidemiological evaluation of this cohort of patients with IPF in the UK confirmed that there was an increased OR of 1.52 for patients taking any antidepressant at the time of diagnosis IPF and an OR of 1.50 for patients taking antidepressants within 4 years of the diagnosis. There was no association with any individual antidepressant or class of antidepressants.160
The clinical course of IPF has been an ongoing subject of debate. Newer data suggest that the progressive decline in respiratory function occurs at a variable rate. Clinical predictors of disease progression and mortality include decline in FVC, decline in diffusion capacity for carbon monoxide (DLCO), arterial oxygen desaturation at rest while breathing room air161 and frequent hospitalizations.162 Increased mortality has been associated with severe respiratory physiologic derangements such as a FVC < 50% or a DLCO < 45%.163 Furthermore, the extent of granulation tissue or fibrosis and degree of ‘young connective tissue’ on histological specimens correlates well with the severity of physiologic derangements VC and DLCO as well as mortality.164,165 King et al. proposed a clinical-radiographical-physiologic (CRP) scoring system to better characterize patients. After a systematic approach to variables suspected to correlate with progression of disease and mortality, the CRP score was developed and included age, smoking status, presence of clubbing, presence of pulmonary hypertension, degree of interstitial opacities, percentage decrease in TLC and partial pressure of arterial oxygen (PaO2) at maximal exercise. Based on their data, 5-year predicted survival for a score of 20 was 89% and <1% for a score of 90.154
In addition to the variable rate of decline in respiratory function, IPF progression is also punctuated by acute exacerbations of disease. These exacerbations may represent a distinct histopathological process from the underlying pathology of UIP. The incidence of acute exacerbations is reported to be between 4.8% and 14.3% over a 9- to 19-month period of follow-up.162,166,167 Prognosis during the exacerbations was worse with a multifocal diffuse pattern of ground glass when compared with isolated peripheral involvement.168–170 It is not currently possible to predict which patients with IPF will experience acute exacerbations, and the mortality rates from acute exacerbations are high. Therefore, many have advocated for early referral for lung transplantation after the initial diagnosis of IPF in order to avoid early mortality.162
The aetiology of acute exacerbations remains unclear, but numerous theories have been postulated. These theories include ongoing exposure to environmental agents, ineffective treatment of GERD, viral infections, oxidant injury, genetic predisposition, autoantibodies or immunologic processes, and presence of distinct cytokine/chemokine profiles. On the basis of these theories, there have been some observations that suggest improved outcomes or delayed progression of IPF with regard to several specific risk factors, specifically tobacco use and GERD.
Although there is not a unifying theory on the aetiology of IPF, the progression of IPF and the aetiology of acute exacerbations, a ‘multi-hit’ hypothesis has now been proposed, whereby any number of potential initial lung injuries in susceptible individuals may lead to abnormal wound repair subsequently inciting a cascade of processes that potentiates fibrosis.171 The remainder of this article will focus on the factors that appear to influence the progression of IPF or have been implicated in the pathogenesis of acute exacerbations including the impact of the proposed treatment options.
Bronchoscopy,166,172,173 surgical lung biopsy174–176 and pulmonary resection for lung cancer177–179 have all been implicated in the pathogenesis of IPF exacerbations. It is not clear whether this is related to a direct procedural effect or whether the development of the exacerbation occurs in association with anaesthesia/anaesthetics, use of high concentrations of oxygen, or ventilator associated acute lung injury/hyper-expansion of the contralateral lung. Contrary to the idea that tobacco use increases the likelihood of developing IPF, newer evidence asserts that there are improved outcomes in patients with IPF who are current smokers.135,154 In vitro data suggest that the protective effects of tobacco smoke may be mediated through carbon monoxide.180
There are new data to suggest seasonal variability in the rates of death from pulmonary fibrosis. Average mortality rates for IPF were 5.2% higher in the fall, 17.1% higher in the winter, 12.7% higher in the spring compared with the rates during summer months.181 These findings were persistent even when documented pneumonia was excluded. They parallel mortality data from patients with COPD, exclusive from pneumonia or infection-related death that also show highest mortality in the winter and spring. On the basis of these data alone, it is not possible to exclude the contribution of viruses or other environmental processes.
Viral infections have been implicated in pathogenesis of IPF as potential initiators of epithelial injury as mentioned previously. Their detection in IPF has also been associated with disease severity, and therefore they are also hypothesized to contribute to the progression of IPF. In vitro studies of EBV have shown that infection of alveolar epithelial cells increased active and latent TGF-β, a cytokine known to be responsible for fibroblast differentiation and increased extracellular matrix deposition. Treatment with Gancyclovir reduced the levels of TGF-β.182 To date there have been no clinical trials evaluating the efficacy of Gancyclovir in the treatment of IPF. EBV has also been implicated in modifying the expression of p53, a tumour suppressor protein that suppresses mitogenic factors and is thought to be responsible for overexpression of growth factors that result in abnormal epithelial cell repair thereby potentially exacerbating or causing progression of the disease process.183 The presence of EBV latent membrane protein 1 (LMP1) in patients with EBV DNA in their lung tissue was associated with higher death rates from respiratory failure than in LMP1 negative patients (4/9 vs 1/20).1 This suggests that specific EBV variants, and not necessarily just EBV infection, could be an important contributor to the progression of IPF.
Other viruses have been implicated in the severity and mortality rates of IPF. Patients with IPF and infected with the TT virus had higher mortality at 3 and 4 years post diagnosis.11 Furthermore, the TT virus has also been implicated in the development of lung cancer in patients with IPF, thereby contributing to mortality rates through carcinogenesis.184 Adenovirus E1A has been identified in patients with IPF. While it was no more frequent in IPF than in the general population, the frequency was significantly higher in IPF and collagen vascular disease associated lung disease patients treated with corticosteroids (67% vs 10%).12 Adenovirus E1A has also been shown to increase levels of TGF-β;185 however, there are no current data to support worse outcomes with adenovirus infection.
These findings raise the question of whether viral reactivation after treatment with immunosuppressive agents may contribute to progression of the disease process, particularly when antiviral therapy is seldom used in this patient population. They also raise the question of whether other viral infections are capable of elaborating TGF-β, similar to adenovirus and EBV, and thereby consolidating a theory of viral induced progression of fibrosis. There are no current data on the role of CMV, adenovirus, HHV-7 & 8, or HSV on progression or exacerbations of IPF. The geographical heterogeneity of viral pathogens and questionable sensitivity of viral detection techniques highlight the difficulty in studying this process.
Several cytokines have been associated with more severe physiologic alterations and/or disease progression. These include IL-8, CCL2/MCP-1, CC-chemokine ligand 18 (CCL18), IL-6 and TNF-α. Serum and BALF of IL-8 levels inversely correlate with DLCO, TLC, VC and, PaO2,78 serum and BALF of CCL2/ MCP-1 levels inversely correlated with DLCO and PaO2,,71,186 and CCL18 levels were inversely associated with disease severity and progression.187 In vitro studies evaluating the role of IL-6 in fibrosis suggest that this cytokine arrests normal fibroblast cell cycles but promotes the cell cycle of fibroblasts from IPF suggesting a pro-proliferative role of IL-6 in the progression of fibrosis.188 Specifically, the IL-6 intron 4GG genotype is linked to diminished DLCO suggesting the importance of IL-6 on progression of disease.189 TNF-α levels are also associated with severity of disease. TNF-α levels from BALF of rapid progressors were significantly higher than controls or stable IPF.190.
Surfactant proteins may also play a critical role in the progression of IPF; however, data are conflicting depending on the source of the proteins. BALF SP-A decreases as pulmonary function decreases,25 and when the SP-A/phospholipids ratio was above the median measurements in BALF from IPF patients, 5-year survival was significantly increased.191 However, when SP-A and SP-D were measured in the serum, elevated levels were associated with an increased 3-year mortality.192 The divergent results of these markers of disease may have to do with leakage of surfactant proteins from the lung tissue into the bloodstream as postulated by Takahashi et al.192 Nevertheless, both BALF and serum levels of surfactant proteins may have a role in predicting outcomes for patients with pulmonary fibrosis.
Some growth factors may be protective from the development of fibrosis. BALF levels of vascular endothelial growth factor (VEGF) were higher in patients with preserved lung function and diminished in patients with poor lung function.186 It is unclear whether VEGF is pro-fibrotic, or rather pro-angiogenic and apoptotic. Further investigation is required to determine whether elevated VEGF may be an indicator of an early phase of disease and diminished VEGF may be an indicator of later phases of disease. Keritinocyte growth factor (KGF) has been shown to be integral to the process of alveolar type II cell division and triggers mitosis in vivo and in vitro.193,194 KGF may also be protective from lung injury. Intratracheal KGF administration protected from acid lung injury, hypoxia, bleomycin and radiation in rats.193,195 Less is known about the role of KGF in human disease.
A genetic predisposition to pulmonary fibrosis may be essential for the initiation of the fibrotic process. Gene expression profiles may eventually help to define various types of pulmonary fibrosis, which may subsequently have implications on whether the pulmonary fibrosis is likely to progress and may also direct treatment. For example, hypersensitivity pneumonitis has a gene expression profile associated with inflammation, T-cell activation and immune responses whereas the profile found in IPF is more notable for epithelial cell and myofibroblast activation.196 The genetic alterations found in IPF add insight to the heterogeneity of this disease, and may eventually help determine focused individualized therapy options.
One of the current enigmas facing the clinical management of IPF is how to determine which patients will have a rapid deterioration in their lung function. While many patients with IPF will have an insidious onset and progression of symptoms, a subset of patients appears to have a more rapidly progressive course. Pulmonary function test (PFT) and BAL data have been compared with histopathological and biomolecular evaluation in an attempt to identify patients at risk for rapid progression. In the lung tissue of rapidly progressive IPF, numerous genes involved in tissue repair and fibrosis were overexpressed. Immunohistochemistry specifically identified two genes of interest expressed by alveolar epithelial cells, adenosine-2B receptor and prominin-1/CD133. Matrix metalloproteinase-9 was also elevated in BALF of rapid progressors. Generally, a subgroup of mostly male smokers had a rapid progression of disease and a discernable gene expression pattern.197 This study is in contrast to other epidemiologic studies suggesting that while smoking increases the risk of IPF, current smokers have a more gradual decline in lung function. Serial analysis of gene expression has also identified surfactant protein A1 and members of a MAPKinase pathway as markers of more rapid progression.198 Information on genetic determinants of the rate of progression from these studies may help in predicting disease progression at the time of diagnosis in IPF patients. Cytokine gene polymorphisms have been associated with the severity of disease when compared with high-resolution CT scores. Specifically, IL-4, IL-4 receptor alpha (IL-4RA), IL-1RA and IL-12 genes were associated with the severity of disease.199
While numerous autoantibodies have been identified in patients, relatively few have been evaluated for correlation with disease severity or have correlated with disease progression and/or severity. Antibodies to IL-1α were increased in five of 11 patients with rapidly progressive IPF on the first day of hospitalization and in all patients on day 21.108 The levels of anti-IL1α increased over the hospitalization. It remains unclear whether IL-1α is specific for IPF or acute exacerbations of IPF. Anitbodies to collagen I and III were inversely correlated with disease duration.100 This correlation was more significant for type III collagen antibodies. Furthermore, antibodies to type II procollagen correlated with the ability of IPF BALF to stimulate fibroblast proliferation in vitro, suggesting a mechanism by which auto-immunity may promote progression of fibrosis.101 An antibody to Annexin 1 was detected in 47% and 53% of serum and BALF respectively from patients with an acute exacerbation of IPF.200 Annexin 1, after cleavage of the N terminus, was also found to induce a proliferative response in CD4-positive T cells. From these data, it is not clear whether this antigen is only important in the process of an acute exacerbation, or whether it is also related to the underlying pathogenesis of UIP/IPF. AntiTh/To antibodies found in ANA positive IPF patients had no impact on survival.94 Similarly, no association was found between disease severity and presence of antitopoisomerase antibodies.102
Circulating fibrocytes are hypothesized to play a role in the development of IPF and other forms of pulmonary and airway fibrosis. Their role in this process has been clarified by identification of the trafficking mechanisms. CXCL12, an important chemotactic factor for wound repair, and its receptor CXCR4 are critical for the recruitment of fibrocytes to the lung interstitium. In mouse models of fibrosis, antibodies to CXCL12 abrogated the ability of fibrocytes to localize to the lung, and in turn, there was a decrease in the extent of inflammation and fibrosis.132 Furthermore, alteration of CXCR4 expression on fibrocytes also decreased their ability to localize in the lung after bleomycin lung injury. The expression of CXCR4 on fibrocytes was enhanced by hypoxia and PDGF.201 CCR2 knock out mice were also unable to recruit fibrocytes to areas of FITC lung injury, and this decrease in fibrocytes correlated with the decrease in lung fibrosis. Furthermore, transplantation with CCR2+/+ bone-marrow restored the ability of fibrocytes to localize to the areas of lung injury and restored the susceptibility to fibrosis.131 These animal models suggest that modulating fibrocyte recruitment may prove an important target for future therapeutic options.
Several small observational series have shown that circulating fibrocyte percentages were elevated in patients with IPF and were particularly high during an acute exacerbation of IPF.202 The levels of fibrocytes have been shown to decrease after resolution of the exacerbation.203 It is not know whether the fibrocytes are responsible for the exacerbation or whether this is merely a compensatory mechanism to promote wound healing; however, elevated levels of circulating fibrocytes in patients without an acute exacerbation correlated with mortality.203
Beeh et al. described lower levels of glutathione in induced sputum and plasma of patients with IPF. The levels of glutathione (GSH) in the sputum correlated with percentage decrease in VC but did not correlate with percentage decrease in DLCO. There was a trend towards association with disease duration.150 Supplementation of oral N-acetylcysteine significantly increased levels of GSH in BALF, and while levels of GSH increased in the epithelial lining fluid compartment of BALF, they did not reach statistical significance.204 Intravenous and aerosolized N-acetyl-l-cysteine (NAC) has been administered with notable transient increases in BALF and ELF GSH levels.205,206 Initial trials of NAC supplementation were encouraging, suggesting a role of antioxidants in the treatment of IPF. The IFIGENIA trial evaluated a primary endpoint of decline in VC and DLCO at 12 months after the addition of NAC to prednisone and azathioprine. There was a slower decline in the DLCO and VC, albeit marginal clinically significant differences.207 To date, NAC has not been studied as a sole treatment regimen; however, future IPF net studies are planned to investigate this anti-oxidant agent further and better clarify the results of the IFIGENIA study.
To date there are no clearly effective treatments for IPF. Numerous treatment regimens have been investigated including corticosteroids, both alone and in combination with other immunosuppressants, antioxidants, Interferon gamma, Bosentan and Pirfenidone. Some of these agents have shown minor clinical benefits; however, none of the available therapies halts the progression or provides a mortality benefit.
Traditional therapies for IPF have included steroids, either alone or in combination with other agents. Corticosteroids have been compared with cytotoxic agents such as azathioprine or cyclophospahmide. Overall, there was a lower rate of decline in FVC in both patients with FVC > 70% and <70% who received cytotoxic agents and this corresponded to a survival benefit.208 More benefit was seen in patients with an FVC > 70%, suggesting that an early intervention was beneficial; however, the rate of infections was significantly higher in the cytotoxic therapy group. Despite these encouraging data, future prospective studies evaluating the effects of cyclophosphamide and prednisone in early stages of IPF failed to show a benefit from the cytoxic agent in terms of preventing physiologic decline or prolonging time to death.209,210 This conclusion is confirmed by the observation that the progression of fibrosis in the native lung of patients with IPF following single lung transplant was not abrogated by a combination of azathioprine, cyclosporine A and prednisone.211 One notable deficit of the trials utilizing azathioprine is a true placebo arm; however, prednisone plus azathioprine has been compared with prednisone plus placebo in a randomized controlled trial of 27 patients with newly diagnosed IPF. This trial showed that there were no significant changes in lung function; however, after correction for age there was a marginally significant survival benefit.212 No study has shown similar benefits, and the azathioprine plus prednisone arm in the IFIGENIA trial had an FVC decline similar to the placebo groups in the interferon and Pirfenidone trials mentioned below. The IFIGENIA trial did also show a slower rate decline in the DLCO and VC measured at 12 months.207 Follow-up studies are planned to either confirm or refute these data.
Pirfenidone is a novel anti-inflammatory/anti-fibrotic agent that was a hopeful pharmacologic therapy for IPF. Pirfenidone did not demonstrate a difference in the primary endpoint, oxygen saturation nadir on a 6-min walk, over the 9-month follow-up, but it did appear to decrease rates of acute exacerbations.167 A follow-up randomized controlled trial (RCT) of Pirfenidone in IPF has closed but the results have not yet been published. The BUILD-1 study was a randomized placebo-controlled trial of Bosentan in IPF. There was no statistical difference in the primary endpoint of 12-month follow-up 6MWD; however, there was a trend towards lower incidence of disease progression and improved quality of life scores.213
The initial trial of gamma-interferon included 18 patients with IPF who had not responded to corticosteroids or other immunosuppressants who were randomized to receive either prednisolone alone or prednisolone plus interferon gamma. The results suggested that disease progression was halted, evidenced by increases in TLC, and in both resting and exercise PaO2.214 However, the results of this initial study were brought into question by King et al. citing concerns for the lack of disease severity and progression, possibility of alternative diagnoses and lack of clinically significant outcomes, despite their statistical significance.215 A subsequent larger multi-centre RCT of patients with IPF who had not responded to corticosteroids were randomized to interferon (IFN)-γ or placebo. This study showed no significant differences in the primary outcome of progression-free survival or the secondary outcomes of pulmonary function or quality of life.216 However, subgroup analysis suggested a mortality benefit for patients with an FVC > 55% and a DLCO > 30%.
A small case series of GERD treatment in IPF suggested that clinical stabilization of IPF could be achieved with adequate control of GERD, and that in this small group of subjects, acute exacerbations of IPF were associated with poor compliance with GERD treatment or recrudescence of GERD217.
In conclusion, a variety of factors have been identified in animal models of pulmonary fibrosis and in patients with IPF. Many of these factors are likely important in both the initiation and progression of this disease, but not all of these factors may be present in every patient with IPF. This highlights the heterogeneity of IPF and how difficult this disease study has been to study. The challenge remains to develop diagnostic strategies that may help determine which insults and pathways are most important in a given patient so that therapeutic interventions can be ‘personalized’. Some of these strategies are discussed in the review, non-invasive biomarkers in pulmonary fibrosis218 (Prasse A, Müller-Quernheim J. Noninvasive biomarkers in pulmonary fibrosis. Respirology 2009 in press), published in an earlier volume of this journal. Identification of biomarkers of disease activity and genetic susceptibilities to specific inciting agents or factors will be important to the ultimate goal of disease prevention.
This work was supported by NIH grants: HL75432, HL088325, HL095402, ES 01247, T32 HL66988, T32 ES07026, The Connor Fund and The Simon Fund. R.M.K. was supported by a Buswell Fellowship.