Genetic Variation in CCL18 Gene Influences CCL18

Journal of Clinical Medicine

Article Genetic Variation in CCL18 Gene Inﬂuences CCL18 Expression and Correlates with Survival in Idiopathic Pulmonary Fibrosis: Part A

Ivo A. Wiertz 1, Soﬁa A. Moll 1, Benjamin Seeliger 2 , Nicole P. Barlo 1, Joanne J. van der Vis 3, Nicoline M. Korthagen 1, Ger T. Rijkers 3, Henk J.T. Ruven 4, Jan C. Grutters 1,5, 2,6, , 1, , Antje Prasse * † and Coline H.M. van Moorsel * † 1 Centre for Interstitial Lung Diseases, Department of Pulmonology, St. Antonius Hospital, 3435 CM Nieuwegein, The Netherlands; [email protected] (I.A.W.); [email protected] (S.A.M.); [email protected] (N.P.B.); [email protected] (N.M.K.); [email protected] (J.C.G.) 2 Department of Respiratory Medicine, Hannover Medical School and Biomedical Research in End-stage and Obstructive Lung Disease Hannover, German Lung Research Center (DZL), 30265 Hannover, Germany; [email protected] 3 Department of Medical Microbiology and Immunology, St. Antonius Hospital, 3435 CM Nieuwegein, The Netherlands; [email protected] (J.J.v.d.V.); [email protected] (G.T.R.) 4 Department of Clinical Chemistry, St. Antonius Hospital, 3435 CM Nieuwegein, The Netherlands; [email protected] 5 Division Heart & Lungs, University Medical Center Utrecht, 3584 CX Utrecht, The Netherlands 6 Fraunhofer Institute for Toxicology and Experimental Medicine, 30625 Hannover, Germany * Correspondence: [email protected] (A.P.); [email protected] (C.H.M.v.M.) Contributed equally to this work. † Received: 25 May 2020; Accepted: 19 June 2020; Published: 21 June 2020

Abstract: Idiopathic pulmonary fibrosis (IPF) is a progressive fibrotic disease, characterized by fibroblast proliferation and extracellular matrix deposition. CC-chemokine ligand 18 (CCL18) upregulates the production of collagen by lung fibroblasts and is associated with mortality. This study was designed to evaluate the influence of single nucleotide polymorphisms (SNPs) in the CCL18 gene on CCL18 expression and survival in IPF. Serum CCL18 levels and four SNPs in the CCL18 gene were analyzed in 77 Dutch IPF patients and 349 healthy controls (HCs). CCL18 mRNA expression was analyzed in peripheral blood mononuclear cells (PBMCs) from 18 healthy subjects. Survival analysis was conducted, dependent on CCL18-levels and -genotypes and validated in two German IPF cohorts (Part B). IPF patients demonstrated significantly higher serum CCL18 levels than the healthy controls (p < 0.001). Both in IPF patients and HCs, serum CCL18 levels were influenced by rs2015086 C > T genotype, with the highest CCL18-levels with the presence of the C-allele. Constitutive CCL18 mRNA-expression in PBMCs was significantly increased with the C-allele and correlated with serum CCL18-levels. In IPF, high serum levels correlated with decreased survival (p = 0.02). Survival was worse with the CT-genotype compared to the TT genotype (p = 0.01). Concluding, genetic variability in the CCL18-gene accounts for differences in CCL18 mRNA-expression and serum-levels and influences survival in IPF.

Keywords: idiopathic pulmonary fibrosis; chemokine; CCL18; single nucleotide polymorphism; survival

J. Clin. Med. 2020, 9, 1940; doi:10.3390/jcm9061940 www.mdpi.com/journal/jcm J. Clin. Med. 2020, 9, 1940 2 of 11

1. Introduction Idiopathic pulmonary fibrosis (IPF) is a progressive fibrotic disease of the lung parenchyma, characterized by fibroblast proliferation and extracellular matrix deposition [1]. New therapeutic agents, such as nintedanib and pirfenidone, demonstrated a deceleration of disease progression [2,3], but overall survival in IPF remains drastically impaired despite these agents [4]. There is substantial inter-individual difference in the clinical course of the disease, ranging from rapid decline to periods of relative stability for many years [1,5]. To predict the disease course, an increasing number of studies investigated the use of several biomarkers in IPF, including data from multicentric randomized trials [6–12]. CC-chemokine ligand 18 (CCL18) is the biomarker most consistently associated with outcomes in IPF and has been studied among several interstitial lung diseases [13–19]. A clear relationship has been demonstrated between the elevated serum levels of CCL18 and clinical outcomes in IPF patients, including survival [16] and acute exacerbation rate [20], and has been confirmed in data from two randomized controlled trials [13]. Apart from IPF, elevated serum CCL18 levels reflected pulmonary fibrosis activity [21] and correlated with death or the progression of pulmonary disease [15,17,22] in patients with systemic sclerosis (SSc)-associated ILD. CCL18 is predominantly expressed by alveolar macrophages and occurs at relatively high levels in lung tissue [23]. In response to CCL18, lung fibroblasts from healthy adults showed an increased expression of collagen mRNA [24]. Furthermore, it was demonstrated that alveolar macrophages from patients with pulmonary fibrosis show an alternatively activated phenotype, which up-regulates the production of collagen by lung fibroblasts through the production of CCL18 [14]. As fibroblast contact and exposure to collagen increases spontaneous CCL18 production by alveolar macrophages, a positive feedback loop was suggested that maintains fibrosis. The gene encoding CCL18 is small, positioned at the q arm of chromosome 17 and consist of three exons with a number of single nucleotide polymorphisms (SNPs), including the SNPs rs2015086 and rs712040, present in the region. In the macrophages of subjects with the A > G genotype of the rs2015086 SNP, a threefold higher gene expression was found compared to those with the A/A genotype [25,26]. We hypothesized that genetic variation in the CCL18 gene might be associated with increased CCL18 expression and may predispose to an unfavorable prognosis in subjects with IPF.

2. Experimental Section

2.1. Patients and Clinical Data A retrospective cohort study was performed in an IPF cohort from the Netherlands. IPF patients diagnosed at St. Antonius Interstitial Lung disease Center of Excellence, Nieuwegein, The Netherlands between 1998 and 2007 were assigned to the derivation cohort (Part A, presented herein). Diagnoses were reviewed and patients were included if the 2011 ATS/ERS criteria were met [27]. Medical records were retrieved to determine survival status, cause of death and baseline characteristics, including age, sex, percentage of predicted forced vital capacity (%FVC) and diﬀusion capacity for carbon monoxide (%DLCO) were recorded. The ﬁndings were then validated in two independent prospectively-recruited German IPF cohorts (presented in Part B of this work). The study protocol was approved by the local Ethical Committee of the St. Antonius Hospital (registration number R05-08A) and all of the subjects gave written informed consent.

2.2. Blood Sampling Serum and blood for DNA extraction were collected at diagnosis. All of the serum samples of patients with the CC and CT genotype were analyzed and, additionally, a random sample from patients with TT were analyzed. For details on DNA analyses, we refer to the supplement. J. Clin. Med. 2020, 9, 1940 3 of 11

2.3. Serum CCL18 Measurement by Monoplex Bead Array CCL18 levels were analyzed using a monoplex suspension bead array system in the derivation cohort. In the validation cohort an ELISA test was used to determine CCL18 levels, as described earlier in other articles [16]. For further details on CCL18 analysis, we refer to the supplement.

2.4. Single Nucleotide Polymorphism (SNP) Genotyping and mRNA Expression Analysis Patients were genotyped for multiple SNPs. Two SNPs, with presumed functionality in the promotor region [26], were genotyped (rs712040, rs2015086). In addition, two haplotype tagging SNPs (rs712042, rs712044) were selected to cover genetic variability in the CCL18 gene. The expressions of CCL18 mRNA in peripheral blood mononuclear cells (PBMCs) from 18 healthy donors were analyzed by quantitative RT-PCR ampliﬁcation. Please refer to the supplementary chapter for further details on genotyping and mRNA expression analysis.

2.5. Statistical Analysis Genotypes were tested for Hardy–Weinberg equilibrium using the website (https://ihg.gsf.de/ cgi-bin/hw/hwa1.pl). Linkage disequilibrium (r2) was calculated using the computer program Haploview 4.1, Broad Institute at Massachusetts Institute of Technology at Harvard University, MA, USA [28]. Haplotypes were determined using Phase v2.1 (Department of Human Genetics, University of Chicago, IL, USA) [29]. The data are presented as medians and interquartile ranges (IQR). Statistical comparisons were made with the use of the Mann–Whitney U test for two groups or Kruskal–Wallis for more than two groups. Receiver operating curves (ROC) were determined to define the optimal cut-off point for distinct serum CCL18 concentrations with the highest predictive accuracy for one year survival. In addition, the Kaplan–Meier method was used for survival analyses. For the analysis of correlation, log-transformation was used to reach near normal distribution. The correlation between mRNA expression and serum levels was assessed using Pearson’s correlation coefficients. Statistical analysis was performed using IBM SPSS statistics software for Windows (version 22.0; IBM, Armonk, NY, USA) and Graphpad PRISM 5 (GraphPad Software, La Jolla, CA, USA) and RStudio version 1.2.5033 (RStudio Inc, Boston, MA, USA). Statistical significance was considered at a value of p < 0.05.

3. Results

3.1. Derivation Cohort A total of 77 IPF patients were included in this study: 58 male, 19 female, median age 61.4 years [IQR 54.1–71.6]). No significant differences were demonstrated in the baseline characteristics between carriers of the different genotypes in the derivation cohort, as shown in Table1. All of the patients were derived from a pre-antifibrotic treatment cohort.

Table 1. Baseline characteristics in IPF patients divided by CCL18 rs2015086 genotype in the derivation cohort.

Characteristics Derivation Cohort (n = 77) rs2015086 Genotype All (n = 77) CT (n = 18) TT (n = 59) p Male, n (%) 58 (75) 11 (61) 47 (79) 0.200 Age, years (IQR) 61.4 (54–72) 61.3 (53–75) 62.8 (52–74) 0.900 Former or active smoker, n (%) 59 (76) 11 (61) 48 (81) 0.950 Baseline %FVC predicted (IQR) 75.7 (62–87) 73.7 (52–88) 75.7 (63–89) 0.700 Baseline %DLCO predicted (IQR) 42.5 (33–56) 40.4 (32–64) 47 (31–60) 0.300 IQR (interquartile range); %FVC (Percent of predicted forced vital capacity); %DLCO (Percent of predicted diﬀusion capacity for carbon monoxide with single breath). J. Clin. Med. 2020, 9, x FOR PEER REVIEW 4 of 10

The genotypes and allele carrier frequencies in the 77 IPF patients of the derivation cohort and 349 controls are summarized in Table 2. The healthy controls and IPF patients were in Hardy– Weinberg equilibrium for all polymorphisms. A comparison of the SNPs in the CCL18 gene revealed no significant differences in allele frequencies between the IPF patients and controls. The following SNPs, including the two SNPs with presumed functionality, showed strong linkage disequilibrium (LD): rs712042, rs2015086 and rs712040; 76 < r2 < 0.90, as shown in Figure 1. Additionally, based on four SNPs, only three haplotypes were constructed with a frequency > 5%. Haplotype frequencies were not significantly different between the IPF patients and healthy controls. Although rs712042 showed the strongest LD, a presumably functional SNP was preferred for further evaluation. Therefore, the rs2015068 polymorphism was selected for analysis.

Table 2. Allele carrier and genotype frequencies in patients with IPF and healthy controls.

IPF Healthy Controls Polymorphism Allele and Genotype (n = 77) (n = 349) * J. Clin. Med. 2020rs712040, 9, 1940 C 16 (10%) 86 (12%) 4 of 11 T 138 (90%) 612 (88%) CC 0 (0%) 7 (2%) 3.2. CCL18 Genotypes and Allele Carrier Frequencies CT 16 (21%) 72 (21%) TT 61 (79%) 270 (77%) DNA was present for 77 patients in the derivation cohort and for 349 healthy subjects (139 male rs2015086 * C 18 (12%) 101 (15%) (40%), 210 female (40%), median age 39.4 years, [IQR 28.3–49.1]). All of the controls and 71 of 77 IPF T 136 (88%) 593 (85%) patients (93%) were of European descent. CC Five IPF patients (6%) 0 (0%) were of North American 10 (3%) descent and one (1%) of North African descent. Differences CT in ethnical background 18 (23%) were not 81 statistically (23%) different between the healthy controls and IPF patients TT (p = 0.184). 59 (77%) 256 (74%) The genotypesrs712042 and allele carrier frequencies A in the 77 136 IPF (88%) patients of the derivation 597 (86%) cohort and 349 controls are summarized in Table2. The G healthy controls and 18 (12%) IPF patients were 101 in (14%) Hardy–Weinberg equilibrium for all polymorphisms. A comparison AA of the SNPs 59 in (77%) the CCL18 gene revealed258 (74%) no significant differences in allele frequencies between AG the IPF patients and 18 controls. (23%) The following 81 (23%) SNPs, including the two SNPs with presumed functionality, GG showed strong 0 linkage(0%) disequilibrium 10 (3%) (LD): rs712042, rs712044 A 97 (63%) 480 (69%) rs2015086 and rs712040; 76 < r2 < 0.90, as shown in Figure1. Additionally, based on four SNPs, G 57 (37%) 218 (31%) only three haplotypes were constructed with a frequency > 5%. Haplotype frequencies were not AA 33 (43%) 172 (49%) significantly diff erent between the IPF AG patients and healthy 31 (40%) controls. Although 136 (39%)rs712042 showed the strongest LD, a presumably functional GG SNP was preferred 13 (17%) for further evaluation. 41 (12%) Therefore, the* healthyrs2015068 controls:polymorphism n = 347, due to was missing selected genotypes for analysis. in two controls. No significant differences were observed.

Figure 1.1. LinkageLinkage Disequilibrium Disequilibrium (LD) (LD) map map of fourof four SNPs SNPs in the in CCL18 the CCL18 gene: rs712040gene: rs712040(location: (location: CCL18 promotor),CCL18 promotor),rs2015086 rs2015086(location: (location: CCL18 promotor), CCL18 promotor),rs712042 rs712042(location: (location: CCL18 intron), CCL18rs712044 intron), (location:rs712044 (location:CCL18 intron). CCL18 The intron). dark squares The dark represent squares high represen r2 valuest high and ther2 values triangle and represents the triangle a haplotype represents block. a haplotype block. Table 2. Allele carrier and genotype frequencies in patients with IPF and healthy controls.

Polymorphism Allele and Genotype IPF (n = 77) Healthy Controls (n = 349) * rs712040 C 16 (10%) 86 (12%) T 138 (90%) 612 (88%) CC 0 (0%) 7 (2%) CT 16 (21%) 72 (21%) TT 61 (79%) 270 (77%) rs2015086 * C 18 (12%) 101 (15%) T 136 (88%) 593 (85%) CC 0 (0%) 10 (3%) CT 18 (23%) 81 (23%) TT 59 (77%) 256 (74%) rs712042 A 136 (88%) 597 (86%) G 18 (12%) 101 (14%) AA 59 (77%) 258 (74%) AG 18 (23%) 81 (23%) GG 0 (0%) 10 (3%) J. Clin. Med. 2020, 9, x FOR PEER REVIEW 5 of 10 J. Clin. Med. 2020, 9, 1940 5 of 11 3.3. Serum CCL18 Levels and CCL18 Genotypes Serum at the time of diagnosis was available from 61 of 77 patients in the derivation cohort Table 2. Cont. (79%). The characteristics of patients with available serum did not differ from the total group, as shown in SupplementaryPolymorphism TableAllele S1. and The Genotype serum levels IPF (weren = 77) further Healthy measured Controls in (n =a349) selection * of 204 healthy controlsrs712044 (86 male (42%), 118 female A (58%), median 97 (63%) age 40.1 years 480[IQR (69%) 29.5 – 51.0]) who were enriched for the presence of the minorG allele rs2015086. 57 (37%) The characteristics 218 of (31%) these healthy controls AA 33 (43%) 172 (49%) did not differ from the total group ofAG controls (p ranging 31 (40%) 0.768–0.999). 136Analysis (39%) of the derivation cohort showed that serum CCL18 levelsGG were significantly 13 (17%) higher in the IPF 41 patients (12%) (645 ng/mL [IQR 393 – *847]) healthy compared controls: n =with347, duethe tohealthy missing genotypescontrols in(185 two ng/mL controls. [IQR No significant 123-272]), differences p < 0.0001, were observed. as shown in Figure 2A. The serum CCL18 levels (median 642 ng/mL [IQR 429-943] in eight patients who received 3.3.low Serumdose corticosteroids CCL18 Levels and were CCL18 not significantly Genotypes different from those of the IPF patients who did not receive immunosuppressive therapy (median 652 ng/mL; IQR 400-856; p = 0.880) at the time of sampling.Serum at the time of diagnosis was available from 61 of 77 patients in the derivation cohort (79%). The characteristicsIn the healthy ofcontrols, patients significant with available differences serum in did serum not di CCL18ffer from levels the totalwere group, observed as shown between in Supplementarythe carriers and Tablenon-carriers S1. The of serum the C-allele levels were of the further rs2015086 measured polymorphism; in a selection TT of 151 204 ng/mL healthy (IQR controls 109- (86224), male CT / (42%), CC 239 118 ng/mL female (IQR (58%), 152-328), median (p age< 0.0001), 40.1 years as shown [IQR in 29.5 Figure – 51.0]) 2B. who No significant were enriched differences for the presencewere found of the in minorCCL18 allele levels rs2015086. between Thethe characteristicsCT-group (241 of ng/mL; these healthy IQR 156-327) controls and did CC-group not differ from(212 theng/mL; total IQR group 137-483; of controls p = 0.834). (p ranging 0.768–0.999). Analysis of the derivation cohort showed that serumPronounced CCL18 levels differences were significantly in the serum higher CCL18 in the IPFlevels patients were (645observed ng/mL between [IQR 393 genotypes – 847]) compared of the withrs2015086 the healthy polymorphism controls (185in the ng IPF/mL patients [IQR 123-272]), of the derivationp < 0.0001, cohort; as shown TT 585 in ng/mL Figure2 (IQRA. The 340 serum –793) CCL18and CT levels 817 ng/mL (median (IQR 642 681 ng /–mL 1278), [IQR p 429-943]= 0.002, as in shown eight patients in Figure who 2C. received No differences low dose were corticosteroids present in werethe baseline not significantly characteristics different between fromthose the patients of the IPF with patients CT and who TT did genotypes not receive of rs2015086, immunosuppressive as shown therapyin Table (median1. 652 ng/mL; IQR 400-856; p = 0.880) at the time of sampling.

Figure 2. Serum CCL18 levels in healthy controls and patientspatients with idiopathic pulmonary ﬁbrosisfibrosis (IPF) in thethe derivationderivation cohort (A), dependingdepending on rs2015086 genotype in healthy controls (B), and IPF patients (C).).

3.4. CCL18In the healthyGenotypes controls, and mRNA significant Expression differences in serum CCL18 levels were observed between the carriers and non-carriers of the C-allele of the rs2015086 polymorphism; TT 151 ng/mL (IQR 109-224), CT / CCThe 239expression ng/mL (IQR of CCL18 152-328), mRNA (p < in0.0001), PBMCs as was shown analyzed in Figure in 182B. healthy No significant controls. di Sixfferences subjects were had foundgenotype in CCL18CT for levelsthe rs2015086 between SNP the and CT-group 12 subjects (241 had ng/mL; TT. IQRThe subjects 156-327) with and CC-groupthe CT genotype (212 ng had/mL; a −5 −5 −5 IQRfour-fold 137-483; higherp = gene0.834). expression (3.0 × 10 ; [IQR 1.8 × 10 − 7.7 × 10 ]) than the subjects with TT (7.4 −6 −6 −5 × 10 Pronounced [IQR 1.1 × 10 di −ff1.8erences × 10 ], in p the= 0.007), serum as CCL18shown in levels Figure were 3A. observed CCL18 mRNA between expression genotypes correlated of the rs2015086significantly polymorphism with serum CCL18 in the levels IPF patients (r = 0.73, of p the = 0.002), derivation as shown cohort; in TTFigure 585 3B. ng/ mLAnalysis (IQR of 340 CCL18 –793) andmRNA CT 817expression ng/mL (IQRfor the 681 rs712040, – 1278), prs712042= 0.002, and as shown rs712044 in Figure SNPs,2 C.respectively, No di fferences showed were similar, present but in non-significant higher gene expression in the CT-genotypes compared with the TT-genotypes. Weak

J. Clin. Med. 2020, 9, 1940 6 of 11 the baseline characteristics between the patients with CT and TT genotypes of rs2015086, as shown in Table1.

3.4. CCL18 Genotypes and mRNA Expression The expression of CCL18 mRNA in PBMCs was analyzed in 18 healthy controls. Six subjects had genotype CT for the rs2015086 SNP and 12 subjects had TT. The subjects with the CT genotype had a four-fold higher gene expression (3.0 10 5; [IQR 1.8 10 5 7.7 10 5]) than the subjects with TT × − × − − × − (7.4 10 6 [IQR 1.1 10 6 1.8 10 5], p = 0.007), as shown in Figure3A. CCL18 mRNA expression × − × − − × − correlated significantly with serum CCL18 levels (r = 0.73, p = 0.002), as shown in Figure3B. Analysis of CCL18J. Clin. Med. mRNA 2020, 9 expression, x FOR PEER for REVIEW the rs712040, rs712042 and rs712044 SNPs, respectively, showed similar,6 of 10 but non-significant higher gene expression in the CT-genotypes compared with the TT-genotypes. Weakto moderate, to moderate, non-significant non-significant correlations correlations between between the theCCL18CCL18 mRNAmRNA and and serum serum CCL18 CCL18 levels levels were werefound found as well as well for for the the rs712040,rs712040, rs712042 rs712042 andand rs712044rs712044 SNPs,SNPs, respectively. respectively.

Figure 3. (A) mRNA expression of CCL18 in PBMCs from 18 healthy controls, expressed as the number of CCL18Figure transcripts3. (A) mRNA per copyexpression of β-actin, of CCL18 according in PBMCs to rs2015086 from 18genotype. healthy (controls,B) Scatterplot expressed showing as the thenumber correlation of CCL18 between transcripts serum CCL18 per copy levels of and β-actin, the number according of CCL18 to rs2015086mRNA transcriptsgenotype. ( perB) Scatterplot copy of β-actin.showing Values the oncorrelation the X and between Y-axis represent serum CCL18 log-transformed levels and values.the number of CCL18 mRNA transcripts per copy of β-actin. Values on the X and Y-axis represent log-transformed values. 3.5. Progression of Disease and Survival in IPF Patients 3.5.Progression Progression of of disease Disease wasand evaluatedSurvival in for IPF the Patients IPF patients, based on FVC change and/or the clinical suspicionProgression of an acute of exacerbation. disease was evaluated Of the 77 patients,for the IP aF total patients, of nine based patients on FVC (13.2%) change had aand/or clinical the suspicionclinical ofsuspicion an acute of exacerbation. an acute exacerbation. A FVC decline Of the of more77 patients, than 10% a total after of one nine year patients was considered (13.2%) had as a FVCclinical progression. suspicion The of FVC an acute values exacerbation. at one-year follow-upA FVC decline were available of more forthan 41 10% out ofafter 77 IPFone patients. year was The median FVC change after one year was 4.5% (IQR 9.4–2.3). Altogether, 18 patients (44%) showed considered as FVC progression. The FVC− values at one-year− follow-up were available for 41 out of 77 a progressionIPF patients. of The disease. median A FVC trend change towards after higher one year baseline was −4.5% CCL18 (IQR levels −9.4–2.3). (652 ng Altogether,/mL; IQR 505-791)18 patients was(44%) observed showed in subjectsa progression with aof progression disease. A trend of disease towards compared higher withbaseline those CCL18 without levels progression (652 ng/mL; (592.8IQR ng 505-791)/mL; IQR was 328-853; observedp = in0.699). subjects No with significant a progression differences of disease were found compared for the with progression those without of diseaseprogression between (592.8 patients ng/mL; with IQR CT 328 and-853; TT genotypesp = 0.699). ofNors2015086 significant(p =differences0.342). The were frequencies found for of the acuteprogression exacerbations of disease did not between differ between patients patients with CT with and the TT CT-genotype genotypes (14.3%)of rs2015086 and TT-genotype (p = 0.342). ofThe rs2015086frequencies(13.0%; of acutep = 0.896). exacerbations did not differ between patients with the CT-genotype (14.3%) and TT-genotypeThe median of survival rs2015086 inthe (13.0%; derivation p = 0.896). cohort was 35 months (95% CI 21.1–48.7). Within the study period,The 50 out median of 77 survival IPF patients in the died derivation (65%), cohort and one was patient 35 months was lost(95% to CI follow-up. 21.1–48.7). ROCWithin analyses the study showedperiod, that 50 theout highest of 77 IPF area patients under thedied curve (65%), (AUC) and wasone calculatedpatient was for lost a serum to follow-up. CCL18 concentration ROC analyses of 500showed ng/mL that (AuC the highest= 0.72). area According under the to thiscurve cut-o (AUC)ff level, was patientscalculated were for categorizeda serum CCL18 as having concentration high (serumof 500 CCL18 ng/mL> (AuC500 ng = /0.72).mL) or According low levels to (serum this cut-of CCL18f level,< 500 patients ng/mL). were The categorized median survival as having in the high (serum CCL18 > 500 ng/mL) or low levels (serum CCL18 < 500 ng/mL). The median survival in the CCL18low-group was 50.4 months (95% CI 31.9–68.9) and differed significantly from that of 27.6 months (95% CI 8.1–47.0) in the CCL18high-group (p = 0.02), as shown in Figure 4A. Survival was also analyzed for dependency on the CCL18 genotype. Patients with the rs2015086 CT genotype showed a significantly worse survival (median 14.3 months (95% CI 0.0–35.9)) compared to the TT genotype (median 37.2 months (15.4–58.9), p = 0.01), as shown in Figure 4B. Patients were censored from the survival analysis if they were alive at end of the follow-up (n = 15) or had received lung transplantation (n = 11). Censored patients were genotyped CT (n = 4) and TT (n = 22). The survival rates were also analyzed in three groups based on a combination of the serum CCL18 level and genotype: CCL18low/TT group, median survival 50.4 months (95% CI 25.4–75.4); CCL18high/TT group, median survival 37.2 months (95% CI 13.1–61.3); CCL18high/CT group, median survival 14.3 months (95% CI 1.4–27.2) (p = 0.03), as shown in Figure 4C. The p-value was calculated

J. Clin. Med. 2020, 9, 1940 7 of 11 J. Clin. Med. 2020, 9, x FOR PEER REVIEW 7 of 10 viaCCL18 the lowLog-group Rank was Test 50.4 using months Kaplan–Meier (95% CI 31.9–68.9) curves and. There diﬀ eredwere signiﬁcantly no patients from with that low of CCL18 27.6 months and genotype(95% CI 8.1–47.0) CT. in the CCL18high-group (p = 0.02), as shown in Figure4A.

Figure 4. Kaplan–Meier curves for survival in patients with idiopathic pulmonary fibrosis, depending Figure 4. Kaplan–Meier curves for survival in patients with idiopathic pulmonary fibrosis, depending on a serum CCL18 level cut-off of 500 ng/mL (A); rs2015086-genotype (B), and a combination of on a serum CCL18 level cut-off of 500 ng/mL (A); rs2015086-genotype (B), and a combination of CCL18 cut-off and rs2015086-genotype (C). P-values were calculated via the Log Rank Test using CCL18 cut-off and rs2015086-genotype (C). P-values were calculated via the Log Rank Test using Kaplan–Meier curves. Kaplan–Meier curves. Survival was also analyzed for dependency on the CCL18 genotype. Patients with the rs2015086 4. Discussion CT genotype showed a significantly worse survival (median 14.3 months (95% CI 0.0–35.9)) compared to theCCL18 TT genotype protein (medianlevels are 37.2 known months as a (15.4–58.9),promising biomarkerp = 0.01), as for shown IPF and in Figurethis study4B. Patients showed were that thecensored results from are thedependent survival on analysis the CCL18 if they genotype. were alive The at endrs2015086 of the follow-upCT polymorphism (n = 15) or in had the received CCL18 genelung contributes transplantation to inter-individual (n = 11). Censored differences patients in were healthy genotyped controls, CT with (n = individuals4) and TT ( ncarrying= 22). the C- allele Thehaving survival the highest rates wereCCL18 also mRNA analyzed and protein in three expression. groups based A similar on a combinationgenotypic effect of theon serum serum CCL18CCL18 levels level andwas genotype:observed in CCL18 patientslow /withTT group, IPF, even median though survival the mean 50.4 serum months levels (95% showed CI 25.4–75.4); a 3.5- foldCCL18 increasehigh/TT group,compared median to the survival healthy 37.2 controls months. Both (95% the CI 13.1–61.3);elevated serum CCL18 CCL18high/CT levelsgroup, and median CT genotypessurvival 14.3 were months related (95% to CIa significantly 1.4–27.2) (p =diminished0.03), as shown long-term in Figure survival4C. The in p-valueIPF. Patients was calculated with the worstvia the survival Log Rank rate Teston the using basis Kaplan–Meier of high serum curves. CCL18 Therelevels werecould no be patientssubdivided with into low intermediate CCL18 and andgenotype worse CT.survival rates according to genotype. Serum CCL18 concentrations reflect pulmonary fibrotic activity in patients with idiopathic interstitial4. Discussion pneumonias (IIPs) and systemic sclerosis with pulmonary involvement [15,21]. Prasse et al. demonstratedCCL18 protein that levels increased are known serum as CCL18 a promising levels biomarker were associated for IPF with and this increased study showedshort-term that (24 the monthsresults arefollow-up) dependent mortality on the inCCL18 IPF patientsgenotype. [16]. The Inrs2015086 our study,CT we polymorphism independently in confirmed the CCL18 thesegene resultscontributes and added to inter-individual to this finding di thefferences predictive in healthy value of controls, serum CCL18 with individuals for long-term carrying survival. the Further, C-allele wehaving showed the highest that serum CCL18 CCL18 mRNA levels and proteinwere genoty expression.pe dependent. A similar genotypicSubjects with effect the on CT serum genotype CCL18 displayedlevels was higher observed constitutive in patients seru withm IPF,CCL18 even levels. though The the CT-genotype mean serum levelsof rs2015086 showed caused a 3.5-fold a four-fold increase highercompared mRNA to the expression healthy controls.in PBMCs Both from the healthy elevated controls. serum The CCL18 influence levels of and genotypes CT genotypes on mRNA were expressionrelated to a in significantly IPF has not diminished been investigated. long-term Intere survivalstingly, in IPF. Hägg Patients et al. with described the worst that survival patients rate with on caroticthe basis artery of high plaques serum and CCL18 the CT levels genotype could of be rs2015086 subdivided had into a three-fold intermediate higher and gene worse expression survival level rates inaccording macrophages to genotype. than subjects with the TT genotype [25]. This is in the same order of magnitude as our resultsSerum and, CCL18 with concentrationsthat, both the genotype–mRN reflect pulmonaryA correlation fibrotic and activity the protein–survival in patients with correlation idiopathic haveinterstitial been demonstrated pneumonias (IIPs) twice and independently. systemic sclerosis with pulmonary involvement [15,21]. Prasse et al., demonstratedAt presentation, that increased patients serum with CCL18 the levelsrs2015086 were CT associated genotype with did increased not show short-term any significant (24 months differencesfollow-up) in mortality demographics in IPF patients or lung function [16]. In our para study,meters we compared independently to patien confirmedts with the these TT resultsgenotype, and asadded shown to thisin Table finding 1. theWe predictive found an value association of serum between CCL18 forthe long-termrs2015086 survival. polymorphism Further, and we showedserum CCL18that serum levels CCL18 in both levels the were controls genotype and dependent.patients. Besides Subjects that, with one the may CT genotypequestion displayedwhether higher higher constitutiveconstitutive CCL18 serum levels CCL18 predispose levels. The to CT-genotypefibrotic disease. of rs2015086In order tocaused investigate a four-fold whether higher the carriage mRNA ofexpression the C-allele in predisposes PBMCs from to healthy IPF, we controls. compared The allele influence frequencies of genotypes between on the mRNA cases expressionand controls in and IPF foundhas not no been significant investigated. differences, Interestingly, even th Häggough et we al. describedhad 80% thatpower patients to detect with an carotic OR ≥ artery 2.1 under plaques a dominantand the CT gene genotype model. of Thers2015086 absencehad of an a three-fold association higher shows gene that expression the carriage level of the in macrophages C-allele does thannot significantly predispose to IPF, however, due to limited sample size, small predisposing effects may still exist.

J. Clin. Med. 2020, 9, 1940 8 of 11 subjects with the TT genotype [25]. This is in the same order of magnitude as our results and, with that, both the genotype–mRNA correlation and the protein–survival correlation have been demonstrated twice independently. At presentation, patients with the rs2015086 CT genotype did not show any significant differences in demographics or lung function parameters compared to patients with the TT genotype, as shown in Table1. We found an association between the rs2015086 polymorphism and serum CCL18 levels in both the controls and patients. Besides that, one may question whether higher constitutive CCL18 levels predispose to fibrotic disease. In order to investigate whether the carriage of the C-allele predisposes to IPF, we compared allele frequencies between the cases and controls and found no significant differences, even though we had 80% power to detect an OR 2.1 under a dominant gene ≥ model. The absence of an association shows that the carriage of the C-allele does not significantly predispose to IPF, however, due to limited sample size, small predisposing effects may still exist. Alveolar macrophages are the main source of CCL18 in the lungs and show an alternatively activated phenotype in IPF [14]. Fibroblast contact and exposure to collagen increases CCL18 production by alveolar macrophages, and these macrophages up-regulate collagen production by lung fibroblasts via the production of CCL18. As such, we hypothesize that an increase in CCL18 does not precede disease but occurs during disease, and that the degree to which CCL18 increases is dependent on the presence of the C-allele. In the search for a biomarker to predict prognosis in IPF, a great number of studies have focused on proteins in serum and BALF. This study is the first to show a genetic polymorphism correlating with serum biomarker levels and with disease course in IPF patients. Genotyping IPF patients for the rs2015086 SNP in the CCL18 gene may, therefore, add substantial information to the interpretation of serum CCL18 levels with regard to the prediction of the disease course. This study has some limitations related to the retrospective nature and relatively small sample size of the IPF cohort. Furthermore, the inclusion period started before 2011 and thereby, patients were selected before the era of new antifibrotic drugs, such as pirfenidone and nintedanib. We could not determine the potential negative effects of immunosuppressive therapy. The difference in age and sex between the IPF patients and healthy controls may have influenced the results. However, with this approach we were able to evaluate a long follow-up period to determine the long-term effect of the biomarker. To address these limitations, in Part B of this work we prospectively validated these findings in two independent German cohorts, one pre-antifibrotic and one with the majority of patients receiving anti-fibrotic therapy. As serum CCL18 levels increase in IPF and influence the disease course, it can be hypothesized that the rs2015086 polymorphism may show similar effects in other fibrotic lung diseases. CCL18 expression is increased in patients with systemic sclerosis and in hypersensitivity pneumonitis [13,17,21,30]. Morbidity and mortality in these diseases are mainly caused by pulmonary fibrosis. Both diseases show a subset of patients who develop a phenotype in which progressive pulmonary fibrosis is the major cause of death. Further research is needed to investigate whether genetic variation in the CCL18 gene influences serum levels and disease course in systemic sclerosis and hypersensitivity pneumonitis. Carriers of the CT genotype are at a disadvantage in terms of higher CCL18 levels and diminished prognosis. Interrupting the positive feedback loop by blocking CCL18 might be an interesting therapeutic intervention. IPF is a relentlessly progressive disease and therapy is not curative and, in general, does not stabilize the disease [3,31,32]. As increased CCL18 levels stimulate fibroblasts to produce collagen, inhibiting CCL18 activity may directly inhibit fibrogenesis. Patients with the CT genotype may especially benefit from a CCL18 blockade as they have the highest serum CCL18 levels.

5. Conclusions In conclusion, we showed that genetic variability in the CCL18 gene accounts for significant differences in CCL18 mRNA expression and serum levels, and was shown to have a modifying role over the course of IPF. Our findings emphasize the value of serum CCL18 as a prognostic marker J. Clin. Med. 2020, 9, 1940 9 of 11 for IPF. Moreover, we confirmed in a replication cohort that future studies concerning CCL18 should take into account that mRNA and protein expression are influenced by genetic polymorphisms in the CCL18 gene. The findings of this study are validated in Part B of this work.

Supplementary Materials: The following are available online at http://www.mdpi.com/2077-0383/9/6/1940/s1, File S1: supplementary chapter; Table S1: characteristics of all IPF patients compared with IPF patients with available serum CCL18. Author Contributions: Conceptualization, N.P.B., G.T.R., A.P. and C.H.M.v.M.; data curation, S.A.M., B.S. and J.J.v.d.V.; formal analysis, I.A.W. and N.P.B.; funding acquisition, J.C.G. and C.H.M.v.M.; investigation, I.A.W., S.A.M., N.P.B., J.J.v.d.V. and N.M.K.; methodology, I.A.W. and C.H.M.v.M.; project administration, S.A.M. and C.H.M.v.M.; resources, J.C.G., A.P. and C.H.M.v.M.; software, J.J.v.d.V. and H.J.T.R.; supervision, J.C.G., A.P. and C.H.M.v.M.; visualization, I.A.W., S.A.M. and B.S.; writing—original draft, I.A.W., N.P.B. and C.H.M.v.M.; writing—review and editing, S.A.M., B.S., J.C.G., A.P. and C.H.M.v.M. All authors have read and agreed to the published version of the manuscript. Funding: ZonMW TopZorg St Antonius Care grant nr 842002001 and Research grant nr 842002003. Acknowledgments: The authors acknowledge Danielle Hijdra and Karin Kazemier for their laboratory assistance. Conﬂicts of Interest: The authors declare no conﬂict of interest.

References

1. Lederer, D.J.; Martinez, F.J. Idiopathic Pulmonary Fibrosis. N. Engl. J. Med. 2018, 378, 1811–1823. [CrossRef] 2. King, T.E.; Bradford, W.Z.; Castro-Bernardini, S.; Fagan, E.A.; Glaspole, I.; Glassberg, M.K.; Gorina, E.; Hopkins, P.M.; Kardatzke, D.; Lancaster, L.; et al. A phase 3 trial of pirfenidone in patients with idiopathic pulmonary fibrosis. N. Engl. J. Med. 2014, 370, 2083–2092. [CrossRef] 3. Richeldi, L.; Du Bois, R.M.; Raghu, G.; Azuma, A.; Brown, K.K.; Costabel, U.; Cottin, V.; Flaherty, K.R.; Hansell, D.M.; Inoue, Y.; et al. Efficacy and safety of nintedanib in idiopathic pulmonary fibrosis. N. Engl. J. Med. 2014, 370, 2071–2082. [CrossRef] 4. Somogyi, V.; Chaudhuri, N.; Torrisi, S.E.; Kahn, N.; Müller, V.; Kreuter, M. The therapy of idiopathic pulmonary fibrosis: What is next? Eur. Respir. Rev. 2019, 28, 190021. [CrossRef] 5. Ley, B.; Collard, H.R.; King, T.E. Clinical course and prediction of survival in idiopathic pulmonary fibrosis. Am. J. Respir. Crit. Care Med. 2011, 183, 431–440. [CrossRef][PubMed] 6. Moeller, A.; Gilpin, S.E.; Ask, K.; Cox, G.; Cook, D.; Gauldie, J.; Margetts, P.J.; Farkas, L.; Dobranowski, J.; Boylan, C.; et al. Circulating fibrocytes are an indicator of poor prognosis in idiopathic pulmonary fibrosis. Am. J. Respir. Crit. Care Med. 2009, 179, 588–594. [CrossRef][PubMed] 7. Maher, T.M.; Stowasser, S.; Nishioka, Y.; White, E.S.; Cottin, V.; Noth, I.; Selman, M.; Rohr, K.B.; Michael, A.; Ittrich, C.; et al. Biomarkers of extracellular matrix turnover in patients with idiopathic pulmonary fibrosis given nintedanib (INMARK study): A randomised, placebo-controlled study. Lancet Respir. Med. 2019, 7, 771–779. [CrossRef] 8. Maher, T.M.; Oballa, E.; Simpson, J.K.; Porte, J.; Habgood, A.; Fahy, W.A.; Flynn, A.; Molyneaux, P.L.; Braybrooke, R.; Divyateja, H.; et al. An epithelial biomarker signature for idiopathic pulmonary fibrosis: An analysis from the multicentre PROFILE cohort study. Lancet Respir. Med. 2017, 5, 946–955. [CrossRef] 9. Kinder, B.W.; Brown, K.K.; McCormack, F.X.; Ix, J.H.; Kervitsky, A.; Schwarz, M.I.; King, T.E. Serum surfactant protein-A is a strong predictor of early mortality in idiopathic pulmonary fibrosis. Chest 2009, 135, 1557–1563. [CrossRef][PubMed] 10. Kinder, B.W.; Brown, K.K.; Schwarz, M.I.; Ix, J.H.; Kervitsky, A.; King, T.E. Baseline BAL neutrophilia predicts early mortality in idiopathic pulmonary fibrosis. Chest 2008, 133, 226–232. [CrossRef] 11. Barlo, N.P.; van Moorsel, C.H.M.; Ruven, H.J.T.; Zanen, P.; van den Bosch, J.M.M.; Grutters, J.C. Surfactant protein-D predicts survival in patients with idiopathic pulmonary fibrosis. Sarcoidosis Vasc. Diffus. Lung Dis. 2009, 26, 155–161. 12. Greene, K.E.; King, T.E.; Kuroki, Y.; Bucher-Bartelson, B.; Hunninghake, G.W.; Newman, L.S.; Nagae, H.; Mason, R.J. Serum surfactant proteins-A and -D as biomarkers in idiopathic pulmonary fibrosis. Eur. Respir. J. 2002, 19, 439–446. [CrossRef][PubMed] J. Clin. Med. 2020, 9, 1940 10 of 11

13. Neighbors, M.; Cabanski, C.R.; Ramalingam, T.R.; Sheng, X.R.; Tew, G.W.; Gu, C.; Jia, G.; Peng, K.; Ray, J.M.; Ley, B.; et al. Prognostic and predictive biomarkers for patients with idiopathic pulmonary fibrosis treated with pirfenidone: Post-hoc assessment of the CAPACITY and ASCEND trials. Lancet Respir. Med. 2018, 6, 615–626. [CrossRef] 14. Prasse, A.; Pechkovsky, D.V.; Toews, G.B.; Jungraithmayr, W.; Kollert, F.; Goldmann, T.; Vollmer, E.; Müller-Quernheim, J.; Zissel, G. A vicious circle of alveolar macrophages and fibroblasts perpetuates pulmonary fibrosis via CCL18. Am. J. Respir. Crit. Care Med. 2006, 173, 781–792. [CrossRef] 15. Prasse, A.; Pechkovsky, D.V.; Toews, G.B.; Schäfer, M.; Eggeling, S.; Ludwig, C.; Germann, M.; Kollert, F.; Zissel, G.; Müller-Quernheim, J. CCL18 as an indicator of pulmonary fibrotic activity in idiopathic interstitial pneumonias and systemic sclerosis. Arthritis Rheum. 2007, 56, 1685–1693. [CrossRef] 16. Prasse, A.; Probst, C.; Bargagli, E.; Zissel, G.; Toews, G.B.; Flaherty, K.R.; Olschewski, M.; Rottoli, P.; Müller-Quernheim, J. Serum CC-chemokine ligand 18 concentration predicts outcome in idiopathic pulmonary fibrosis. Am. J. Respir. Crit. Care Med. 2009, 179, 717–723. [CrossRef] 17. Schupp, J.; Becker, M.; Günther, J.; Müller-Quernheim, J.; Riemekasten, G.; Prasse, A. Serum CCL18 is predictive for lung disease progression and mortality in systemic sclerosis. Eur. Respir. J. 2014, 43, 1530–1532. [CrossRef] 18. Pechkovsky, D.V.; Prasse, A.; Kollert, F.; Engel, K.M.Y.; Dentler, J.; Luttmann, W.; Friedrich, K.; Müller-Quernheim, J.; Zissel, G. Alternatively activated alveolar macrophages in pulmonary fibrosis-mediator production and intracellular signal transduction. Clin. Immunol. 2010, 137, 89–101. [CrossRef] 19. Cai, M.; Bonella, F.; He, X.; Sixt, S.U.; Sarria, R.; Guzman, J.; Costabel, U. CCL18 in serum, BAL fluid and alveolar macrophage culture supernatant in interstitial lung diseases. Respir. Med. 2013, 107, 1444–1452. [CrossRef] 20. Schupp, J.C.; Binder, H.; Jäger, B.; Cillis, G.; Zissel, G.; Müller-Quernheim, J.; Prasse, A. Macrophage activation in acute exacerbation of idiopathic pulmonary fibrosis. PLoS ONE 2015, 10, e0116775. [CrossRef] 21. Kodera, M.; Hasegawa, M.; Komura, K.; Yanaba, K.; Takehara, K.; Sato, S. Serum pulmonary and activation-regulated chemokine/CCL18 levels in patients with systemic sclerosis: A sensitive indicator of active pulmonary fibrosis. Arthritis Rheum. 2005, 52, 2889–2896. [CrossRef][PubMed] 22. Tiev, K.P.; Hua-Huy, T.; Kettaneh, A.; Gain, M.; Duong-Quy, S.; Tolédano, C.; Cabane, J.; Dinh-Xuan, A.T. Serum CC chemokine ligand-18 predicts lung disease worsening in systemic sclerosis. Eur. Respir. J. 2011, 38, 1355–1360. [CrossRef][PubMed] 23. Adema, G.J.; Hartgers, F.; Verstraten, R.; de Vries, E.; Marland, G.; Menon, S.; Foster, J.; Xu, Y.; Nooyen, P.; McClanahan, T.; et al. A dendritic-cell-derived C-C chemokine that preferentially attracts naive T cells. Nature 1997, 387, 713–717. [CrossRef][PubMed] 24. Atamas, S.P.; Luzina, I.G.; Choi, J.; Tsymbalyuk, N.; Carbonetti, N.H.; Singh, I.S.; Trojanowska, M.; Jimenez, S.A.; White, B. Pulmonary and activation-regulated chemokine stimulates collagen production in lung fibroblasts. Am. J. Respir. Cell Mol. Biol. 2003, 29, 743–749. [CrossRef] 25. Hägg, D.A.; Olson, F.J.; Kjelldahl, J.; Jernås, M.; Thelle, D.S.; Carlsson, L.M.S.; Fagerberg, B.; Svensson, P.-A. Expression of chemokine (C-C motif) ligand 18 in human macrophages and atherosclerotic plaques. Atherosclerosis 2009, 204, e15–e20. [CrossRef] 26. Modi, W.S.; Lautenberger, J.; An, P.; Scott, K.; Goedert, J.J.; Kirk, G.D.; Buchbinder, S.; Phair, J.; Donfield, S.; O’Brien, S.J.; et al. Genetic variation in the CCL18-CCL3-CCL4 chemokine gene cluster influences HIV Type 1 transmission and AIDS disease progression. Am. J. Hum. Genet. 2006, 79, 120–128. [CrossRef] 27. Raghu, G.; Rochwerg, B.; Zhang, Y.; Garcia, C.A.C.; Azuma, A.; Behr, J.; Brozek, J.L.; Collard, H.R.; Cunningham, W.; Homma, S.; et al. An Official ATS/ERS/JRS/ALAT Clinical Practice Guideline: Treatment of Idiopathic Pulmonary Fibrosis. An Update of the 2011 Clinical Practice Guideline. Am. J. Respir. Crit. Care Med. 2015, 192, e3–e19. [CrossRef] 28. Barrett, J.C.; Fry, B.; Maller, J.; Daly, M.J. Haploview: Analysis and visualization of LD and haplotype maps. Bioinformatics 2005, 21, 263–265. [CrossRef] 29. Stephens, M.; Smith, N.J.; Donnelly, P. A new statistical method for haplotype reconstruction from population data. Am. J. Hum. Genet. 2001, 68, 978–989. [CrossRef] 30. Pardo, A.; Smith, K.M.; Abrams, J.; Coffman, R.; Bustos, M.; McClanahan, T.K.; Grein, J.; Murphy, E.E.; Zlotnik, A.; Selman, M. CCL18/DC-CK-1/PARC up-regulation in hypersensitivity pneumonitis. J. Leukoc. Biol. 2001, 70, 610–616. J. Clin. Med. 2020, 9, 1940 11 of 11

31. Wells, A.U.; Flaherty, K.R.; Brown, K.K.; Inoue, Y.; Devaraj, A.; Richeldi, L.; Moua, T.; Crestani, B.; Wuyts, W.A.; Stowasser, S.; et al. Nintedanib in patients with progressive ﬁbrosing interstitial lung diseases—subgroup analyses by interstitial lung disease diagnosis in the INBUILD trial: A randomised, double-blind, placebo-controlled, parallel-group trial. Lancet Respir. Med. 2020, 8, 453–460. [CrossRef] 32. Flaherty, K.R.; Wells, A.U.; Cottin, V.; Devaraj, A.; Walsh, S.L.F.; Inoue, Y.; Richeldi, L.; Kolb, M.; Tetzlaﬀ, K.; Stowasser, S.; et al. Nintedanib in Progressive Fibrosing Interstitial Lung Diseases. N. Engl. J. Med. 2019, 381, 1718–1727. [CrossRef][PubMed]

© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).