Presence of anti-modified protein antibodies in idiopathic pulmonary fibrosis

Background and Objective: Studies of autoimmunity and anti-citrullinated protein antibodies (ACPA) in idiopathic pulmonary fibrosis (IPF) have been confined to investigations of anti-cyclic citrullinated peptide (anti-CCP) antibodies which utilize synthetic peptides as surrogate markers for in vivo citrullinated antigens. We studied immune activation by analysing the prevalence of in vivo anti-modified protein antibodies (AMPA) in IPF. Methods: We included patients with incident and prevalent IPF ( N = 120), sex and smoking-matched healthy controls (HC) ( N = 120) and patients with RA ( N = 104). Serum (median time: 11 months [Q1 – Q3: 1 – 28 months] from diagnosis) was analysed for presence of antibodies towards native and posttranslational modified (citrullinated [Cit, N = 25]; acetylated [Acet, N = 4] and homocitrullinated [Carb, N = 1]) peptides derived from tenascin (TNC, N = 9), fibrinogen (Fib, N = 11), filaggrin (Fil, N = 5), histone ( N = 8), cathelicidin (LL37, N = 4) and vimentin ( N = 5) using a custom-made peptide microarray. Results: AMPA were more frequent and in increased levels in IPF than in HC (44% vs. 27%, p < 0.01), but less than in RA (44% vs. 79%, p < 0.01). We specifically observed AMPA in IPF towards certain citrullinated, acetylated and carbamylated peptides versus HC: tenascin (Cit (2033) -TNC 2025 – 2040 ; Cit (2197) -TNC 2177 – 2200 ; Cit (2198) -TNC 2177 – 2200 ) , fibrinogen (Cit (38,42) -Fib α 36 – 50 ; Cit (72) -Fib β 60 – 74 ) and filaggrin (Acet-Fil 307 – 324 , Carb-Fil 307 – 324 ). No differences in survival ( p = 0.13) or disease progression ( p = 0.19) between individuals with or without AMPA was observed in IPF. However, patients with incident IPF had better survival if AMPA were present ( p = 0.009). Conclusion: A significant proportion of IPF patients present with specific AMPA in serum. Our results suggest autoimmunity as a possible characteristic for a subgroup of IPF that may affect disease outcome.


INTRODUCTION
Idiopathic pulmonary fibrosis (IPF) and rheumatoid arthritisassociated interstitial lung disease (RA-ILD) share several common characteristics.2][3] Usual interstitial pneumonia (UIP) pattern on high resolution computed tomography (HRCT) is a hallmark in IPF, but is also frequently observed in RA. [4][5][6] These features, but also similarities in factors associated with poor prognosis 6,7 raises the question of shared disease mechanisms and common autoimmune and inflammatory pathways.
Several antibodies implicated in the development of RA and their associations with RA-ILD have been described over the years. 8Detection of autoimmune antibodies constitutes an important tool not only in the diagnosis of RA, but also in the exclusion of pulmonary fibrosis associated with connective tissue disease.A part of the specific RA antibodies target proteins subject to different posttranslational modifications (Anti-modified protein antibodies, AMPA) such as citrullination, 9,10 carbamylation 11 and acetylation. 12utoreactivity towards citrullinated proteins is highly specific for RA and have been found in bronchoalveolar lavage (BAL), and associated with RA-ILD 6,[13][14][15] -in contrast to posttranslational modifications such as citrullination per se that is found in inflamed tissues. 16Citrullinated proteins have, for example, been detected in BAL cells and lung tissue from healthy smokers, 17 patients with IPF, 18,19 RA-ILD 19 and RA. 20Autoreactivity towards carbamylated 21 and acetylated 12 proteins is less specific for RA and not studied systematically in IPF.Studies of early and long-standing RA with ILD have shown associations with the presence and number of ACPA reactivities in sera. 22,23Presence of ACPA and their targeting of citrullinated autoantigens in different cell structures is called the ACPA reactivity repertoire.
Taken together, there is evidence suggesting a possible role of posttranslational modifications of proteins, in particular citrullination in IPF and plausible immune activation with antibody reactivity.5][26][27][28] Anti-CCP test is based on a synthetic cyclic citrullinated peptide and utilized in the clinic for the detection of ACPA when diagnosing RA.Few studies in IPF have investigated antibodies targeting naturally occurring citrullinated proteins and antibodies targeting other posttranslationally modified proteins.Microarray methods that simultaneously detect AMPA have recently been developed and tested in patients with RA. 29 Likewise, no previous studies have determined whether patients with IPF have distinct reactivity repertoire of ACPA and AMPA compared with patients with RA.Exploring and investigating these autoantibody responses may give new clues to the aetiology and the heterogeneity seen in IPF.
With this background, we set out to investigate the presence of ACPA and AMPA in serum from patients with IPF.Furthermore, we explored eventual clinical phenotypes associated with AMPA and their impact on disease progression and survival.

IPF patients
One hundred and twenty patients diagnosed with IPF between 2007 and 2020, enrolled from Karolinska University Hospital, Stockholm, Sweden, and the Swedish IPFregistry were included.Patients were diagnosed with IPF according to national and international guidelines. 30,31A diagnosis of IPF was based on clinical-radiological evaluation in 93% (n = 112), while 7% (n = 8) required a lung biopsy.In our clinic, patients with significant serology or other clinical signs of autoimmune disease are referred to rheumatologists to exclude autoimmune diseases.Serum samples were collected, aliquoted and stored at À70 C within 2 h of sampling until use and obtained at a median time of 11 months (Q1-Q3: 1-28 months) from diagnosis.No additional exclusion criteria were applied.More details are provided in the Supporting Information.

Controls and early RA patients
Serum samples from 120 sex-and smoking matched healthy controls (HC) from the Epidemiological Investigation of Rheumatoid Arthritis (EIRA) cohort were used. 32Likewise, serum samples from 104 patients with recent onset rheumatoid arthritis (RA) with symptom duration <1 year, fulfilling the American Rheumatism Association 1987 classification criteria for RA 33 from the LUng involvement in Rheumatoid Arthritis (LURA) study at Karolinska University Hospital, Stockholm, were also analysed. 34Lung function measures were obtained from spirometry performed 1 week after RA diagnosis.None of the patients with RA were diagnosed with interstitial lung disease.

SUMMARY AT A GLANCE
We show that multiple anti-modified protein antibodies are present in a significant proportion of patients with idiopathic pulmonary fibrosis (IPF).Antibodies against anti-modified peptides were present in patients despite having no antibodies against anti-cyclic citrullinated peptide.The presence of these antibodies may be associated with longer survival in IPF.
The tested peptides derive from suggested citrullinated autoantigens, detected in the RA synovia and in vitro experiments, and are proteins related to extracellular matrix (ECM), tenascin 35 and fibrinogen, 36,37 the epithelial barrier associated filaggrin, 38,39 mesenchymal cytoskeletal vimentin 40 and chromosome associated histones. 41

Statistical analysis
Differences between groups were assessed with Student's t-test and Mann-Whitney U-test, as appropriate.Descriptive statistics are presented as median and quartiles (Q1-Q3).Comparisons of frequency distributions of reactivity between groups were calculated with Pearson's Chi-square or Fischer's exact test when appropriate.Analyses of survival and time to progression grouped by reactivity were performed using Kaplan-Meier curves and comparisons done using the log-rank test.Adjusted Cox proportional hazards were used to examine the predictive value of AMPA status on survival while controlling for potential influential variables such as disease severity and smoking.Statistical analyses were performed in the Graphpad Prism 9 software.Results were considered significant if the p-value was <0.05.
Detailed information on statistical analysis and outcome measures can be found in the Supporting Information.

Patients and controls
Baseline demographics and clinical data are shown in Table 1.Patients with IPF were older than HC and RA and comprised of more ex-smokers than in the RA cohort.More RA patients were females and current smokers compared to IPF and HC.Lung function was significantly lower in IPF compared to RA across all measures.Classification according to Gender-Age-Physiology (GAP) index, showed that approximately half (48%) of IPF patients were classified as GAP-stage 2, 36% as GAP-stage 1 and 15% as GAP-stage 3. Measures of 6MWT performance were available for 98 IPF-patients, with a reported median (Q1-Q3) distance walked of 443 m (377-501) and a lowest oxygen saturation during test of 88% (85%-91%).A majority of patients with IPF had negative ANA (71%, N = 81), negative RF (89%; N = 78), negative anti-CCP (95%; N = 75) and negative ANCA (100%; N = 106).Arterial hypertension was the most common comorbidity among IPF patients (49%), followed by gastroesophageal reflux (33%), other cardiovascular diseases (which includes heart failure and fibrillation, 21%), followed by diabetes (16%) and thyroid-related diseases (11%).
Antibody reactivity to the native counterpart was not increased except for antibody reactivity to Fibβ 36-52 (Table S1 in the Supporting Information).A majority of IPF patients (68%, N = 36) had multiple as opposed to just single AMPA reactivity (Figure S1 in Supporting Information).When IPF patients were divided into three groups with increased number of AMPA reactivities (0, 1-5 and >5 reactivities), the IPF cohort had an increased number of individuals with >5 reactivities compared to HC (12% vs. 3%, p = 0.003).More than half (51%) of the individuals in the RA-cohort had >5 reactivities (p < 0.0001 vs. IPF and HC).No differences between IPF and HC were observed regarding AMPA targeting specific modified vimentin, histone or LL37 peptides (Table S1 in the Supporting Information).RA patients showed increased prevalence and concentrations of AMPA against a total of 19 peptides compared to IPF and 24 peptides compared to HC (Table S1 in the Supporting Information).Evaluation of AMPA reactivities in patients with reported tests of anti-CCP revealed that 41% (N = 31) of patients who had tested negative (N = 75) presented with reactivity against at least one of the tested modified peptides.Similarly, 45% (N = 35) of patients who tested negative for RF (N = 78) showed reactivity.

AMPA and clinical phenotype in IPF
IPF patients with or without AMPA did not differ with respect to age, sex, smoking exposure, lung function, disease severity measured by GAP (Table 2) or prevalence of comorbidities (data not shown).Evaluation of clinical characteristics, comorbidities and demographic data in IPF patients grouped by reactivity showed no differences between the groups (Table S2 in the Supporting Information).Clinical characteristics (Table S3 in the Supporting Information) and prevalence of comorbidities (data not shown) did not differ between patients with or without AMPA in the incident IPF subcohort.However, the proportion of patients who were alive was higher among AMPA positive compared to AMPA negative patients (80% vs. 52%, p = 0.02).Treatment with antifibrotics were similar between AMPA negative and positive patients in the full cohort (Table 2) and the incident subcohort (Table S3 in the Supporting Information).

AMPA, disease progression and survival
During a median observation period of 35 months (Q1-Q3: 21-51), approximately half of the patients with IPF had died (N = 55, 48%) or been transplanted (N = 10, 8%).There was no significant difference in survival time from serum collection between IPF patients with and without AMPA (median survival: 51 vs. 39 months, respectively, p = 0.13) (Figure 2) nor when patients were grouped by the number of reactivities (data not shown) (p = 0.20).Similar results were obtained in analyses of time to progression, where no differences on the basis of reactivity were observed (p = 0.19, Figure S2 in the Supporting Information and p = 0.33 when grouped by number of reactivities (data not shown), respectively).Additional analysis of the incident IPF subcohort showed improved survival in those with presence of AMPA compared to AMPA negative patients (median survival: not reached vs. 42 months, respectively, p = 0.009; Figure 3A).When the cohort was separated into an increasing number of reactivities, a similar pattern of survival was obtained (Figure 3B).Although not statistically significant, IPF patients with no antibodies had a trend towards increased mortality compared to patients with 1-5 reactivities (logrank p = 0.062), but significantly increased mortality compared to patients with >5 reactivities (log-rank p = 0.034).
Time to progression did not differ significantly when grouped by presence/absence of AMPA (p = 0.12) or by the Abbreviations: AMPA, anti-modified protein antibodies; DLCO%, diffusing capacity of carbon monoxide, % predicted; FEV 1 %, forced expiratory volume in 1 s, % predicted; FVC%, forced vital capacity, % predicted; GAP, Gender-Age-Physiology; TLC%, total lung capacity,% predicted; 6MWT, 6-minute walking test.number of reactivities (Figure S3A,B in the Supporting Information), but a similar trend of better prognosis was observed.Multivariate Cox analysis including GAP stage, smoking status and antifibrotic therapy were performed in the full and incident cohort.Presence of AMPA were associated with a trend towards increased survival in the full cohort (HR: 0.63, 95% CI: 0.37-1.05,p = 0.08, Table S4 in the Supporting Information) and the incident cohort (HR: 0.41, 95% CI: 0.15-0.99,p = 0.06, Table S4 in the Supporting Information).

DISCUSSION
This study contributes to the limited literature on AMPA reactivities, antibody reactivity to several posttranslational modified peptides such as ACPA in patients with IPF.Our results demonstrate that almost half of the patients with IPF present with a differentiated immune activation and antibodies to posttranslationally modified proteins with a significant proportion (12%) having multiple reactivities (>5 reactivities).This includes reactivities to epitopes from tenascin, fibrinogen and filaggrin.The pattern of AMPA reactivities were different in both their prevalence and levels compared to healthy matched controls and patients with early RA where there was an overall pattern of reactivities in IPF targeting citrullinated, carbamylated and acetylated peptides.This is different from the AMPA reactivity pattern in RA which is predominantly citrulline specific, commonly targeting citrullinated vimentin and citrullinated filaggrin and at higher levels. 42Although the differentiation in AMPA reactivity does not seem to be reflected in the clinical characteristics between IPF patients, the presence and number of AMPA may be associated with longer survival.Our results demonstrated that more than a third of IPF patients have ACPA despite negative anti-CCP.Several naturally occurring molecules recognized by ACPA have been described, and the tested antibodies in this study are well established, including fibrinogen and filaggrin, 36 vimentin 40 and tenascin. 35Their exact role in the lung are yet to be elucidated and additional mechanistic studies investigating the interaction between these antibodies and peptides and their role in the lung in IPF are warranted.
5][26][27][28] A possible explanation is differences in capturing reactivities to various peptides such as tenascin.Studies investigating the role of antibodies (IgG) such as RF, ANA and particularly anti-CCP and its prognostic importance in IPF are very limited and have shown no significant associations to disease progression and survival. 43Our study, therefore, contributes with important data on the potential role of certain ACPA and AMPA and survival.We found a significantly longer survival in patients with the presence and increasing number of these antibodies in the incident subcohort compared to patients lacking these specific AMPA.Similar trends, although not statistically significant, were observed in analyses of progression in AMPA positive patients.The difference in outcome measures between the full cohort and the incident cohort is likely due to the difference in time between diagnosis and serum collection.With a median time of 30 months from diagnosis to sampling among prevalent patients as opposed to 2 months among incident, the differences in disease duration are likely to result in worse outcome measures and thus bias the analyses of survival.
More recently, Solomon et.al showed that patients with IPF were more likely to be carriers of IgA anti-CCP (21%-25%) as opposed to IgG anti-CCP (55-8%). 44lthough the multiplex assay used in this study only analyses the IgG isotype, our results, suggesting high prevalence of IgG ACPA and IgG AMPA when testing for multiple reactivities and not just CCP, together with those presented by Solomon et al., highlight the potential role of the adaptive immune system and autoimmunity's role in the pathogenesis of IPF.Importantly, it raises intriguing questions that needs to be investigated in larger prospective cohort studies.
Among the limitations, we do not know whether included IPF patients developed a defined connective tissue disease in the following years.This raises interesting questions on the potential of misclassifications of IPF, which theoretically could be a potential explanation to our results of increased prevalence of AMPA in IPF.However, we would argue that the incidence is expected to be very low 45 and hence have a possible minor influence, not enough to explain the almost doubling of prevalence of AMPA in IPF versus HC presented in this study.Although theoretically possible of occult autoimmune disease, our patients are diagnosed according to clinical guidelines and referred to rheumatologists if motivated.However, we cannot completely rule out that patients with an occult autoimmune disease may be included.Moreover, analysing comparative patterns of AMPA reactivity against RA-ILD patients using unbiased methods (such as immunoprecipitation and mass spectrometry) 13 would be a possible way of detecting IPFspecific autoantigens.Another limitation is the absence of specific data on immunosuppressive treatment in the IPF cohort.
In Sweden though, patients with IPF are not treated with high doses of corticosteroids or other immunosuppressants.Overall, there are important strengths of our study: the relatively large cohorts of healthy controls, RA patients and patients with IPF with, in the context of IPF, a novel focus of ACPA reactivities and AMPA.Furthermore, the longitudinal followup of the IPF cohort made it possible to explore and generate hypotheses on the long-term implications on survival and progression based on AMPA reactivity.In summary, IPF patients have increased prevalence and levels of certain AMPA targeting tenascin, fibrinogen and filaggrin presenting in a different pattern compared to healthy controls and RA.Our results suggest autoimmunity in IPF as a possible mechanism for a subgroup of patients and that the presence of AMPA reactivity may be associated with longer survival.

2
Survival time estimates were calculated from the time of serum collection based on the presence or absence of AMPA against RAassociated peptides in the full IPF cohort.

F
I G U R E 3 (A,B) Survival time estimates in incident IPF patients were calculated from the time of serum collection based on the presence or absence of AMPA (A) and grouped according to the number of AMPA reactivities (B) against RA-associated peptides.
Characteristics of patients with idiopathic pulmonary fibrosis (IPF), rheumatoid arthritis (RA) and healthy controls.
T A B L E 2 Characteristics of idiopathic pulmonary fibrosis patients grouped according to the presence of AMPA reactivity.