Molecular Biomarkers in Idiopathic Pulmonary Fibrosis: State of the Art and Future Directions

Idiopathic pulmonary fibrosis (IPF), the most lethal form of interstitial pneumonia of unknown cause, is associated with a specific radiological and histopathological pattern (the so-called “usual interstitial pneumonia” pattern) and has a median survival estimated to be between 3 and 5 years after diagnosis. However, evidence shows that IPF has different clinical phenotypes, which are characterized by a variable disease course over time. At present, the natural history of IPF is unpredictable for individual patients, although some genetic factors and circulating biomarkers have been associated with different prognoses. Since in its early stages, IPF may be asymptomatic, leading to a delayed diagnosis. Two drugs, pirfenidone and nintedanib, have been shown to modify the disease course by slowing down the decline in lung function. It is also known that 5–10% of the IPF patients may be affected by episodes of acute and often fatal decline. The acute worsening of disease is sometimes attributed to identifiable conditions, such as pneumonia or heart failure; but many of these events occur without an identifiable cause. These idiopathic acute worsenings are termed acute exacerbations of IPF. To date, clinical biomarkers, diagnostic, prognostic, and theranostic, are not well characterized. However, they could become useful tools helping facilitate diagnoses, monitoring disease progression and treatment efficacy. The aim of this review is to cover molecular mechanisms underlying IPF and research into new clinical biomarkers, to be utilized in diagnosis and prognosis, even in patients treated with antifibrotic drugs.


Introduction
Idiopathic pulmonary fibrosis (IPF) is a chronic, progressive lung disease [1]. The epidemiology of this disease is not uniform due to data collection methods and classification terms variability among different studies. However, throughout Europe and North America, an incidence between 2.8 and 19 cases per 100,000 people per year has been reported [2][3][4]. IPF primary affects men, older than 50 years (median age at diagnosis is about 65 years) [5][6][7]. The disease course is variable, due to different clinical phenotypes [8,9]. However, the median survival time from diagnosis is 2-4 years [10]. Since in its early stages, IPF may be asymptomatic, leading to a delayed diagnosis. When present, the most frequent symptoms are progressive dyspnoea and cough. The IPF diagnosis is based on the identification of the usual interstitial pneumonia (UIP) pattern, both on histological samples or radiological images, and the exclusion of other known causes of pulmonary fibrosis. Frequently the diagnosis is complex, requiring a multidisciplinary evaluation as recommended by international guidelines [11,12]. At present, two drugs, nintedanib and pirfenidone, which slow the progression of the disease and improve prognosis, are approved for the treatment of IPF [12].
The management of IPF is currently based on clinical data, such as symptoms, lung function tests, and radio-histological patterns, due to a lack of reliable molecular markers. However, the identification of clinical biomarkers, diagnostic, prognostic, and theranostic would allow an evaluation based on underlying pathobiological mechanism of disease, leading to adequate phenotyping of patients in terms of diagnosis, prognosis, and response to therapy. This review aims to cover molecular mechanisms underlying IPF and research into new clinical biomarkers, to be utilized in diagnosis and prognosis, even in patients treated with antifibrotic drugs.

Definition of a Biomarker
Biomarkers are defined as "characteristics that are objectively measured and evaluated as an indicator of normal biologic processes, pathogenic processes or pharmacologic responses to a therapeutic intervention" [13]. At any time during the evaluation of patients affected by a disease, biomarkers can be considered useful tools. Predisposition biomarkers could identify people at risk for eventually develop a disease, diagnostic and prognostic biomarkers integrate the diagnostic process and theranostic biomarkers are a reliable measure of efficacy and safety during treatment. Moreover, biomarkers are frequently used as a surrogate endpoint in clinical trials helping predict clinical benefit based on epidemiologic, therapeutic, pathophysiologic, or other scientific evidence [13]. Currently, 2018 American Thoracic Society (ATS), European Respiratory Society (ERS), Japanese Respiratory Society (JRS), American Latin Thoracic Association (ALAT) strongly recommends not to measure any serum biomarker for the purpose of distinguishing IPF from other interstitial lung diseases (ILD) in patients with newly detected ILD of apparently unknown cause who are clinically suspected of having IPF. Moreover, no guidelines or official statement on prognostic and theranostic biomarkers are available.
A good-quality biomarker should be reproducible, very sensitive, specific, and accurate. It should be validated in large multicentric trials and heterogeneous populations. Moreover, to be used on a large scale, it should be easily available and accessible. Biomarkers detectable on peripheral blood, exhaled breath condensate or broncho-alveolar lavage (BAL) offer an increased range of applications compared with a transbronchial or surgical lung biopsy. Finally, the cost-effectiveness ratio should be acceptable [14].

Molecular Biomarkers in IPF
The development of new molecular biomarkers for IPF is based on two different approaches. The hypothesis-driven method selects new candidate biomarkers a priori based on previous evidence about the disease. In contrast, the unbiased approach utilizes methods from systems biology to screen a large number of candidate biomarkers for their association with the disease. Although the former has the advantage of a strong rationale but lacks efficiency, the latter is more efficient but also burdened by the risk of false discovery [14].
Historically, IPF was considered a chronic inflammatory disorder, gradually leading to fibrosis. However, anti-inflammatory, and immunosuppressive therapy have shown to be ineffective and associated with increased mortality [15]. Up to date, IPF is described as characterized by the interaction of multiple genetic and environmental risk factors, with local micro-injuries to ageing alveolar epithelium. As a consequence, different process such as aberrant epithelial-fibroblast communication, the induction of myofibroblasts and the accumulation of extracellular matrix, lead to remodeling of lung interstitium [16]. Consequently, the most promising biomarkers in IPF are related to alveolar epithelial cell dysfunction, immune dysregulation, fibroproliferation, fibrogenesis, and extracellular matrix remodeling [14].
The main biomarkers analyzed in this review and their possible applications are resumed in Table 1 and Figure 1.

Predisposition Biomarkers
Mechanism or biological pathways linked to disease predisposition are reflected by predisposition biomarkers. They should provide information through inexpensive and non-invasive sampling with high sensitivity, specificity, and predictive value. Using predisposition biomarker, a patient could be address to informative counselling, preventive measures, and early therapy [14].
A common single nucleotide polymorphism (SNP) in the putative promoter region of the mucin 5B (MUC5B) gene (rs35705950), which encoded a glycosylated macromolecular component of mucus, has been associated with familiar pulmonary fibrosis and sporadic pulmonary fibrosis [27][28][29][30][31][32][33][34]. A recent meta-analysis confirmed that the minor T allele is significantly associated with an increased risk of IPF compared with the G allele in an allele dose-dependent manner [35]. However, MUC5B (rs35705950) has been found in 9% of people with interstitial lung abnormalities (ILA), a prevalence way higher than the rate reported for IPF [36]. Although MUC5B promoter polymorphism is a promising predisposition biomarker, it is neither necessary nor sufficient to cause the disease and understanding its role in IPF pathogenesis together with other genetic or environmental factors remain an unmet need.
The telomerase complex is involved in protection of chromosomes from loss of material, catalyzing the addition of repeated DNA sequences in the telomere region [37]. Several proteins contribute to the correct activity of the telomerase complex, including telomerase reverse transcriptase (TERT), dyskerin, telomere binding protein (TIN2), interaction with the telomerase repeat binding factor (TERF1), and the telomerase RNA component (TERC). Moreover, several other proteins contribute to the regulation of telomerase complex [37]. Several variants of the telomerase complex and its regulatory proteins have been associated to pulmonary fibrosis, especially familiar forms [38]. Although the telomerase complex can be evaluated on blood cells (granulocytes or monocytes), and common in IPF patients compared with age matched controls [39], it is globally rare in sporadic IPF and not specific since it has been associated with risk of developing chronic obstructive pulmonary disease (COPD) too [40].
Although the relationship between IPF and immunity is controversial, several components of the immune system have been evaluated as predisposition biomarkers in IPF. Toll-like receptors (TLR), fundamental components of innate immunity, have been implicated in the pathogenesis of IPF. TLR-2 mRNA is overexpressed in IPF patients and has shown pro-fibrotic features in mice. TLR-3 has shown antifibrotic features both in human and mice through downregulation of transforming growth factor beta 1 (TGFB1) and upregulation of prostaglandin E2 (PGE2). The loss-of-function variants (L412F) of TLR-3 lead to enhanced fibrotic responses. Some evidence, mainly in animal models, has been shown also for TLR2/4, TLR9, and TLR4 [41][42][43][44]. Another candidate gene for IPF is ELMOD-2, expressed in alveolar macrophages and type II AECs. A genome wide scan in 6 families with familiar pulmonary fibrosis in Finland showed reduced levels of mRNA expression of ELMOD-2 in IPF patients compared with healthy controls [45].
Predisposition biomarkers help understanding the pathogenesis of IPF and predicting the predisposition and prognosis of the disease. However, to date, none of these biomarkers is completely specific and sensitive for the diagnosis of sporadic pulmonary fibrosis, nor validated in clinical use. In familiar cases of pulmonary fibrosis, a consult with a geneticist and a screening for the most common biomarkers should be proposed to patients.

Diagnostic Biomarkers
Diagnostic biomarkers should reflect the mechanism or biological pathways that distinct IPF from the other ILDs. They should be easy to evaluate and reproducible, such as blood, urine, BALF derived or imaging-based biomarkers. Ideally, they should improve diagnosis, reduce the risk of diagnostic tests, reduce the number of unclassifiable cases, helping discriminate IPF from other ILDs accurately [14]. Several blood proteins have shown some evidence in terms of diagnostic process, however, the use of none of them is recommended by guidelines for the diagnosis of IPF [11].
The detection of surfactant proteins in serum of patients with pulmonary diseases reflects an injury of the alveolar epithelial barrier. SP-A and D have been studied as diagnostic markers in IPF. Although BALF levels of surfactant proteins are reduced both in IPF and other ILDs compared with healthy controls [46,47], serum levels appear to be increased [47][48][49]. Wang et al. conduced a meta-analysis to evaluate the use of SP-A and SP-D for differential diagnosis of IPF. SP-A serum levels appear to be significantly higher in patients with IPF than in patients with non-IPF ILDs, pulmonary infection, and healthy controls, while no differences are found in SP-D serum levels in IPF versus non-IPF ILD patients, although higher than those in pulmonary infection and healthy controls [50]. Recently, SP-B precursor, C-pro-SP-B, has been studied as a new biomarker in serum of patients with different chronic lung diseases including ILDs. The highest serum levels of C-pro-SP-B were detected in the serum of IPF patients being able to differentiate IPF patients from patients with all other pulmonary diseases [51]. Moreover, SP-D levels were significantly elevated in acute exacerbation of IPF compared with stable IPF [52].
Krebs von den Lungen-6 (KL6)/mucin 1 (MUC1) is a glycoprotein expressed on the extracellular surface of type II AECs and bronchiolar epithelial cells in the lung largely studied in ILDs due to its overexpression in affected lung and regenerating type II AECs [53,54]. KL-6 is increased in serum of several ILDs including IPF [47,49,53,[55][56][57] In one study, KL-6 levels in BALF seems to be a specific diagnostic marker in IPF compared with other ILDs [58] while Bennet et al. proved that higher levels of BALF KL-6 are related to a more severe and extended disease [56]. However, since KL-6 reflects AECs damage, it is not specific enough to distinguish IPF from the other ILDs nor alone neither as a part of composite index. However, it could facilitate stratification of severity [55][56][57].
Circulating caspase-cleaved cytoke-ratin-18 (cCK-18) is the cleaved fragment of cytokeratin-18 (CK-18), a cytoskeletal protein found in AECs. Since cCK-18 is produced during apoptosis in response to stress, it has been evaluated as a diagnostic and prognostic marker in one study: cCK18 was significantly elevated in the serum of IPF patients compared with normal controls and patients with other ILDs although it was not associated with prognosis [59].
TLRs have been studied widely for their implication in IPF pathogenesis and predisposition. However, some evidence has highlighted a possible role in diagnosis. Higher levels of TLR in BALF of IPF patients, particularly TRL-7 has been noted. In the same study, TLR also showed different profiles of expression in fibrotic and granulomatous disorders [60].
Metalloproteases (MMP) are another class of proteins widely studied for their role in in the aberrant fibrotic process, but the mechanisms are not completely understood and characterized as it seems they are implied both in deleterious and beneficial effects on the fibrotic process [61] They are a family of zinc-dependent matrixins that participate in extracellular matrix degradation but also in processing and cleaving of different bioactive mediators [62]. In particular, MMP-1 is upregulated in IPF patients compared with controls, and higher levels of MMP-1 has been shown in BALF and in plasma of IPF patients [63,64]. MMP-7 is also upregulated in IPF, with higher serum and BALF levels in patients compared with healthy controls [14,65]. Recently, Bauer et al. analyzed samples from the Bosentan Use in Interstitial Lung Disease (BUILD)-3 trial dosing MMP-7 among other biomarkers. MP-7 protein levels were elevated in IPF patients compared with healthy controls, and MMP-7 levels also increased over time [66]. Although MMP-7 alone is not sufficiently specific to distinguish IPF from other ILDs, if evaluated with other markers of fibrosis it could help differentiate IPF from other ILDs with good accuracy [64,67].
Osteopontin (OPN) is a phosphorylated glycoprotein that work as a mediator of inflammation and wound healing [68]. OPN is overexpressed in IPF lung [68,69] and seems its profibrotic role seems to be related to its ability to enhance fibroblasts migration by cooperating with chemoattractant interleukin 6 (IL-6) [69]. Moreover, OPN also induces upregulation of MMP-7 [70]. Although OPN is increased in serum and BALF of IPF patients [71,72], it is not specific in differentiating IPF from other ILDs [72]. However, as part of a composite index it helps improving diagnostic confidence [67].

Prognostic Biomarkers
Prognostic biomarkers should contribute to quantitative assessment of mechanism or biological pathways relevant to disease progression. They should be repeatable over time without significant risk for patients, such as blood and urine-based biomarkers. Moreover, they should have a low intra-patient, inter-test variability with calibration and discrimination values clearly established. They could be integrated in multiparametric models in order to improve prognostic counselling [14]. Potentially useful prognostic biomarkers have been identified both in genomic variants and blood proteins. With regard for genomic variants mutation in the telomerase complex and MUC5B have been studied as possible prognostic biomarkers. Telomere length has been associated to a worse survival [73] and transplant free survival [74] in IPF patients. Although MUC5B promoter variant (rs35705950) has been associated to an increased risk of developing pulmonary fibrosis, its role in predicting survival is contradictory [75][76][77][78]. Serum levels of SP-A and SP-D have been associated with reduced survival in IPF [46,48,[79][80][81]. The variant in the TOLLIP gene, encoding for an adapter protein, is associated with a worse survival and more rapid disease progression possibly helping stratification at baseline of IPF patients [78].
Several studies highlighted a relationship between elevated values of KL-6 and mortality or progression in IPF [57,[82][83][84][85] although these data have not been confirmed in other studies [86,87]. On the other hand, serial measurements of serum KL-6 concentrations resulted a risk factor for progressive disease and worse prognosis [57,88]. With regard for AE of IPF, serum values of KL-6 resulted higher compared with stable patients and higher values are predictor of onset of AE [52,89,90]. A recent systematic review and meta-analysis suggest that increased values of KL-6 in IPF is a predictor of AE risk, while it seems not to be related with mortality [87].
With regard for immune mediators, the SNP in theTLR3 (L412F) has been associated to increased mortality and accelerated progression in independent cohort [91,92]. Alpha-defensins, small antimicrobial proteins secreted by neutrophils and epithelial cells, has been proposed as a biomarker of AE of IPF. In fact, although alpha-defensins are upregulated in IPF lung, higher levels are detectable in patients with AE of IPF. Moreover, alpha-defensin serum levels were increased in AE IPF compared with stable IPF suggesting their use as biomarkers for AEs [63].
MMPs, in particular MMP-7, have been studied not only as diagnostic biomarkers but they can be useful tools in predicting prognosis and transplant free survival in IPF patients [14,66]. MMP-7 has also been evaluated in several studies in association with other markers of IPF for its diagnostic and prognostic qualities with positive results [67,86,93]. MMP-7 is not the only MMPs that has shown promising results as a diagnostic biomarker. Recently, Todd et al. evaluated the circulating serum levels of MMPs (MMPs 1, 2, 3, 7, 8, 9, 12, and 13) and tissue inhibitors of MMPs (TIMP) (TIMPs 1, 2, and 4) in a cohort of 300 IPF patients from the IPF-PRO Registry, highlighting that MMPs and TIMPs analyzed were all present at higher levels in patients with IPF compared with controls except for TIMP2. MMP8, MMP9, and TIMP1 were the best diagnostic markers for distinguishing patients with IPF from controls. Moreover, MMP7, MMP12, MMP13, and TIMP4 were able to stratify patients for disease severity [94].
The evidence on the prognostic value of OPN is scarce [72,95]. However, interestingly, a recent study showed that OPN serum levels where significantly higher in patients with AE of IPF compared with stable IPF or healthy controls. Moreover, higher levels of OPN were associated to increased mortality in AEs [96]. Periostin, another ECM protein involved in tissue development and wound healing, has been shown part of the pathogenetic process in IPF. Periostin has prognostic values, in fact total periostin can predict both short-term declines of pulmonary function and overall survival in IPF patients. However, total periostin is not specific for IPF. On the contrary, the monomeric periostin form is more specific and can be used not only to predict pulmonary function decline but also to distinguish IPF patients from healthy controls [97].

Therapeutic Biomarkers
Therapeutic biomarkers should provide quantitative assessment or indicate the presence or absence of mechanisms or biological pathways targeted by therapy. Since these biomarkers work as surrogate endpoints, they should be measurable over time with low risks for patients, low intra-patient, inter-test variability, and should improve clinical decision making of therapeutic intervention. Finally, when a therapeutic biomarker is evaluated threshold for change should be established reflecting meaningful therapeutic response [14].
Since many biomarkers have been studied before the introduction of antifibrotic therapy, evidence on their usefulness in monitoring response to therapy is more limited.
Surfactant proteins serum levels have shown potential efficacy as outcomes in IPF therapy. SP-A levels in IPF patients treated with pirfenidone or nintedanib from baseline to 3 and 6 months were found to predict progression [98], while SP-D levels in IPF patients treated with IPF predict disease progression and prognosis [99,100].
Promising data have been shown in several studies on KL-6 and its use in the monitoring of antifibrotic therapy. Response to pirfenidone therapy correlates with changes in serum KL-6 over time in one study [101] Bergantini et al. evaluated serial measurements of serum KL-6 in IPF patients treated with nintedanib, and demonstrated an indirect correlation with forced vital capacity (FVC) percentages and KL-6 values. Moreover, after 1 year of treatment, patients on therapy showed stable FVC percentages and KL-6 levels compared with baseline values [102]. Nakamura et al. did not find any difference in KL-6 serum values in severe IPF patients treated with nintedanib compared with non-severe patients [103].
MMP-7 has recently been studied in with other biomarkers of transplant free survival in a study on 325 patients, 68 of them treated with antifibrotic therapy, to evaluate the role of such biomarkers in patients on antifibrotic therapy. The study revealed that these biomarkers predict differential transplant free survival in patients on antifibrotic therapy but at higher thresholds than in non-treated patients. Moreover, plasma biomarker level generally increases over time in non-treated patients but remain unchanged in patients on antifibrotics [104].

Conclusions
The need of reliable biomarkers is becoming more and more fundamental. The validation of useful and accurate diagnostic markers could reduce uncertainty and the use of invasive procedure. Prognostic and therapeutic markers could help stratify patients based on severity and disease behavior in order to personalize management. Moreover, reliable markers able to predict AEs could implement prevention measures and modify the prognosis of such events, which, to date, is poor. Several molecules have shown potential value as biomarkers in IPF. However, many of them have been evaluated mainly in Asiatic cohorts of patients, where their use is more common. Their accuracy should be confirmed also in Caucasian cohorts in order to routinely apply them in the management of IPF. The use of biomarker index composed by multiple biomarkers already studied separately, with the aim of improve diagnostic accuracy in distinguish IPF from other ILDs or healthy controls is promising, but, for now, has shown controversial results.
Finally, none of these biomarkers have been validated in large clinical trials, which still remain an unmet need. However, as a remark of the importance of biomarkers in IPF, many clinical trials evaluating as primary or secondary outcomes known and new biomarkers, have been conducted (Tables 2 and 3) or are still ongoing (Tables 3 and 4).             Identify a group of genetic loci that play a role in the development of familial interstitial pneumonia and idiopathic interstitial pneumonia.
Develop biomarkers using proteomic and genomic approaches that will facilitate establishing the diagnosis and prognosis of both familial and sporadic forms of idiopathic interstitial pneumonia  Conflicts of Interest: P.F., F.L. and A.P. report personal fees from Roche and Boehringer-Ingelheim, outside the submitted work. All the other authors have no conflicts of interest to declare.