Article Text

Original research
Medical treatments for idiopathic pulmonary fibrosis: a systematic review and network meta-analysis
  1. Tyler Pitre1,
  2. Jasmine Mah2,
  3. Wryan Helmeczi3,
  4. Muhammad Faran Khalid4,
  5. Sonya Cui4,
  6. Melanie Zhang4,
  7. Renata Husnudinov4,
  8. Johnny Su1,
  9. Laura Banfield5,
  10. Brent Guy4,
  11. Jade Coyne4,
  12. Ciaran Scallan1,6,
  13. Martin RJ Kolb1,6,
  14. Aaron Jones7,
  15. Dena Zeraatkar7,8
  1. 1 Division of Internal Medicine, McMaster University, Hamilton, Ontario, Canada
  2. 2 Department of Medicine, Dalhousie University, Halifax, Nova Scotia, Canada
  3. 3 Division of Internal Medicine, University of Ottawa, Ottawa, Ontario, Canada
  4. 4 Michael G. DeGroote School of Medicine, McMaster University, Hamilton, Ontario, Canada
  5. 5 Faculty of Health Sciences, McMaster University, Hamilton, Ontario, Canada
  6. 6 Division of Respirology, St. Joseph's Hospital, Hamilton, Ontario, Canada
  7. 7 Health Evidence Impact and Research, McMaster University, Hamilton, Ontario, Canada
  8. 8 Bioinformatics, Harvard Medical School, Cambridge, Massachusetts, USA
  1. Correspondence to Dr Dena Zeraatkar, Havard Medical School, Harvard University, Cambridge, Massachusetts, USA; Dena_Zeraatkar{at}hms.harvard.edu

Abstract

Background Idiopathic pulmonary fibrosis (IPF) is a respiratory disorder with a poor prognosis. Our objective is to assess the comparative effectiveness of 22 approved or studied IPF drug treatments.

Methods We searched MEDLINE, EMBASE, Cochrane Central Register of Controlled Trials and clinicaltrials.gov from inception to 2 April 2021. We included randomised controlled trials (RCTs) for adult patients with IPF receiving one or more of 22 drug treatments. Pairs of reviewers independently identified randomised trials that compared one or more of the target medical treatments in patients with IPF. We assessed the certainty of evidence using the Grading of Recommendations Assessment, Development, and Evaluation (GRADE) approach for network meta-analysis. We calculated pooled relative risk (RR) ratios and presented direct or network estimates with 95% credibility intervals (95% CI), within the GRADE framework.

Results We identified 48 (10 326 patients) eligible studies for analysis. Nintedanib [RR 0.69 (0.44 to 1.1), pirfenidone [RR 0.63 (0.37 to 1.09); direct estimate), and sildenafil [RR (0.44 (0.16 to 1.09)] probably reduce mortality (all moderate certainty). Nintedanib (2.92% (1.51 to 4.14)), nintedanib+sildenafil (157 mL (–88.35 to 411.12)), pirfenidone (2.47% (–0.1 to 5)), pamrevlumab (4.3% (0.5 to 8.1)) and pentraxin (2.74% (1 to 4.83)) probably reduce decline of overall forced vital capacity (all moderate certainty). Only sildenafil probably reduces acute exacerbation and hospitalisations (moderate certainty). Corticosteroids+azathioprine+N-acetylcysteine increased risk of serious adverse events versus placebo (high certainty).

Conclusion and relevance Future guidelines should consider sildenafil for IPF and further research needs to be done on promising IPF treatments such as pamrevlumab and pentraxin as phase 3 trials are completed.

  • idiopathic pulmonary fibrosis
  • interstitial fibrosis
  • rare lung diseases

Data availability statement

Data are available upon reasonable request.

Statistics from Altmetric.com

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Key messages

What is the key question?

  • For treatment of idiopathic pulmonary fibrosis (IPF), the comparative efficacy of drug treatments is unclear for numerous patient important outcomes and there have been novel treatments that have been previously not compared.

What is the bottom line?

  • Nintedanib, pirfenidone and sildenafil probably reduce mortality in patients with IPF. Pamrevlumab may reduce mortality.

Why read on?

  • Current guidelines will need to address the potential of novel treatments for IPF in light of emerging evidence.

Background

Idiopathic pulmonary fibrosis (IPF) is a progressive interstitial lung disease with an overall poor prognosis. Notwithstanding lung transplantation, only nintedanib and pirfenidone have been proven in large trials to slow reduction in disease progression and overall mortality.1 2

There have been previous network meta-analyses (NMA) that demonstrate the efficacy and safety of commonly studied and proposed treatment for IPF.2 3 However, there have been multiple new proposed treatments, with promising results since the last major review of the literature and appraisal of the evidence, have not been directly compared against each other or older treatments.

The objective of this review is to evaluate the comparative effectiveness and safety of 22 approved and investigational treatments for IPF.

Methods

We registered a protocol on Open Science Framework at https://osf.io/afbhd/, posted the protocol as a preprint, and included the protocol as a supplement (online supplemental appendix 1, p34.).4 This review is reported in accordance with the Preferred Reporting Items for Systematic Review and Meta-Analysis (PRISMA) extension statement for reporting of systematic reviews incorporating NMAs.5 There are no funding sources for this study.

Supplemental material

Search strategy

We searched MEDLINE, EMBASE, CENTRAL and Clinicaltrials.gov from inception to 2 April 2021 (online supplemental appendix 1, p3–6). We did not include unpublished trials.

Eligibility criteria and screening

We included randomised controlled trials (RCTs) that reported on patients 18 years of age or older, diagnosed with IPF and compared one or more of the 22 different treatments (single or combination) against each other or placebo. We included RCTs for adult patients with IPF receiving one or more of 22 drug treatments. We did not limit based on the type of RCT, date, language or number of patients included.

Data management and selection process

Reviewers, working independently and in duplicate using COVIDENCE (an online citation manager), screened titles and abstracts of search records, and subsequently, the full texts of records deemed potentially eligible at the title and abstract screening stage. Reviewers resolved discrepancies by discussion, and if necessary, by third-party adjudication.

Data collection process and data outcomes

We extracted data using a pilot-tested data extraction form in excel, based on the Cochrane data extraction form. We collected data on trial and baseline patient characteristics (eg, country, age, sex) and outcomes of interest.

We reached out to authors for additional information or clarification when necessary. We reached out specifically to one author concerning pentraxin mortality data and did not receive a reply. We did not need to access raw data.

Outcomes

Our outcomes included mortality, reduction in disease progression and serious adverse events (SAE). Disease progression was measured as a mean change in forced vital capacity (FVC) (either change in millilitres (mL) or per cent change predicted), 6 minute walk distance(6-MWD) distance, diffusion capacity of carbon monoxide (DLCO) and number of hospitalisations and acute exacerbations. For adverse events, we assessed SAE and adverse events leading to drug discontinuation. We synthesised outcomes at the longest reported point of follow-up. Definitions for each outcome are found inonline supplemental appendix 1, p38.

Risk of bias in individual studies

We assessed risk of bias independently and in duplicate using a revision of the Cochrane tool for assessing risk of bias in randomised trials (RoB V.2.0).6 We rated each outcome as either at (1) low risk of bias, (2) probably low risk of bias, (3) probably high risk of bias or (4) high risk of bias, across the following domains: bias arising from the randomisation process; bias owing to departures from the intended intervention; bias from missing outcome data; bias in measurement of the outcome; bias in selection of the reported results, including deviations from the registered protocol and bias arising from early termination for benefit. We resolved discrepancies by discussion and, when not possible, with adjudication by a third party. Our risk of bias tool is found in online supplemental appendix 1, p7.

Data synthesis

We summarised the effect of interventions of our dichotomous outcomes (eg, mortality) using relative risks (RR) and corresponding 95% credibility intervals (CIs) as well as absolute risk reduction per 1000 patients. We calculated the baseline risk by calculating the median risk in the placebo arm across trials. For continuous outcomes, we reported mean difference and 95% CIs. For FVC, we analysed trials reporting mean change of per cent predicted and change in millilitre separately. Treatment nodes were grouped based on class of medication (ie, bosentan, ambrisentan, macitentan were grouped as endothelin receptor antagonists). Combination treatments were treated as separate nodes. We combined placebo and standard care into a single node since patients receiving placebo would also be receiving standard care with no other investigational treatments. We excluded treatment nodes for which there was only a single study with no events observed in either arm. We considered all treatment doses as equivalent and multiarm trials with multidose regimens were pooled together. For each comparison that included more than 10 studies, we planned to test for evidence of publication bias using statistical (ie, Egger test) and graphical methods (ie, funnel plots). Given that none of our comparisons included more than 10 studies, we did not test for assessment publication bias.7

For all outcomes, we conducted Bayesian random-effects network meta-analysis, setting vague priors for variance and effect parameters. A Bayesian method produces wider CIs as compared to direct estimates owing to assumed increased network heterogeneity . Therefore, we adjusted for this when rating for imprecision and focused on interpreting the point estimate. A network meta-analysis uses direct and indirect estimates to produce network estimates, which are what is typically reported in the results. Network estimates can be used to compare relative efficacy of drug treatments. This can be done by comparing how the network estimates for each drug treatment compare against placebo or by looking at the head-to-head estimates. We primarily report network estimates versus placebo but provided the head-to-head estimates in online supplemental appendix 3. For direct estimates, we performed an inverse variance random effects meta-analysis, seen in online supplemental appendix 1, p47. We used four Markov chains with 200 000 iterations, a burn-in of 10 000, and a thinning factor of 10. We assessed convergence by examining trace plots and Gelman diagnostics. Node splitting models were used to compare direct and indirect evidence when possible.8 We used STATA V.17 to make our network plots and performed our network meta-analysis using the R package gemtc, V.0.82.9 10

Supplemental material

Certainty of the evidence

We assessed the certainty of the evidence independently and in duplicate using the Grading of Recommendations, Asssesement, Development, and Evaluations (GRADE) approach for network meta-analysis.11 Online supplemental appendix 2 provides more detail on GRADE for those unfamiliar with this type of analysis and is highly recommended for readers. In adherence to the GRADE system of rating the certainty of evidence, we rated the certainty for each comparison and outcome as high, moderate, low or very low, based on considerations of risk of bias, inconsistency, indirectness, publication bias, intransitivity, incoherence (difference between direct and indirect effects) and imprecision.12–20 Intransitivity is the dissimilarity of important factors that may affect the outcome being investigated (ie, effect modifiers) across comparisons and coherence refers to consistency between direct and indirect estimates. We assess for intransitivity by reviewing the distribution of potential effect modifiers across comparisons. We considered disease severity to be a convincing effect modifier (baseline FVC, DLCO, 6-MWD). We considered baseline disease severity to be the only potential effect modifier that may have introduced intransitivity in the network. We did not consider criteria for the diagnosis of IPF to be an effect modifier because although modalities for diagnosis (eg, biopsy, CT scan) have changed, the underlying population has not appreciably changed. This is the standard method of rating the certainty of the evidence for network meta-analysis within the GRADE framework.21 Definitions are found in the cited literature and in online supplemental appendix 2, which includes a detailed description of how we rated the evidence. Importantly, statistical significance is not considered in GRADE and there is a plethora of literature describing the tendency away from statistical significance and instead focus on interpretation of the point estimate.22–25

Supplemental material

We used a minimally contextualised approach when rating imprecision, using the most updated GRADE guidance.17 This is one of the standardised ways of rating the certainty of evidence developed by the GRADE working group.15 We considered a 2% and 5% reduction in mortality and acute exacerbations and hospitalisations, respectively, as minimally important. We considered a 5% threshold for SAE and a 20% threshold for adverse events, leading to drug discontinuation as minimally important. When FVC was reported as per cent predicted, we used a minimal important difference of 2% and when FVC was reported in millilitre, we used 100 mL difference.26 27 For the 6-MWD and DLCO, we considered 35 m and 17% reduction as minimally important.28 29

We presented the results of our NMA for drugs against placebo, with colour-associated ranges of the certainty of evidence, seen in table 2. This is the accepted and most standardised method for presenting the results of NMA; all relevant head-on comparisons are summarised in online supplemental appendix 3. We used the GRADE simple language summary to present the certainty of evidence (high certainty=drug reduces mortality; moderate certainty=drug probably reduces mortality; low certainty=drug may reduce mortality; very low certainty=the effect of drug X on mortality is very uncertain).30

Results

Our search identified 5250 records, 151 of which were screened in full and 48 trials, including 10 326 patients, were eligible. Figure 1 presents the PRISMA flowchart of inclusion of RCTs. The trials recruited patients from over 20 countries. The majority of the trials were multicentred. The mean age of patients was 66.7 years old, and patients were predominantly men. The majority of participants in most of the trials had a smoking history. The baseline FVC per cent predicted was reduced across most studies. Table 1 presents the essential patient characteristics, and online supplemental table 1 has the individual trial characteristics.27 31–65 Online supplemental appendix 3 contains the network estimates for all outcomes. For all outcomes, the networks were connected. Online supplemental appendix 1 p27–33 has the network diagrams. We rated down the certainty of evidence most often for imprecision of the estimate and risk of bias ratings. There was little heterogeneity found in the networks; table 2 presents the I2 for each network. We did not perform a sensitivity analysis based on risk of bias.

Supplemental material

Table 1

Basic characteristics across trials

Table 2

Drug nodes versus placebo, presented in absolute risk reduction per 1000 patients (95% credibility intervals), with the negative sign indicating fewer patients per 1000 and the positive indicating more per 1000

Figure 1

PRISMA flow diagram for identification of studies. PRISMA, Preferred Reporting Items for Systematic Review and Meta-Analysis.

Risk of bias

One-third of trials were at high risk of bias across one or more domains across various outcomes. There are 11 (22.9%) studies at risk of bias for their randomisation process, 15 (31.2%) were at risk of bias for deviations from their intended interventions, 10 (20.8%) were at risk of bias for possible missing data, 6 (12.5%) were at risk for bias for the measurement of the outcome and 6 (12.5%) were at risk for selection of the reported results. Online supplemental appendix 1 p13–26 presents detailed risk of bias assessments and online supplemental appendix 1 p7–12 has our modified ROB V.2.0 tool.

GRADE

We present detailed GRADE ratings for direct and network estimates with each GRADE domain assessed in online supplemental appendix 3. Otherwise, we present how each of our estimates were rated in our summary of findings table as compared with placebo.

Mortality

Forty-seven trials reported on mortality, including 9901 patients with randomised IPF and 915 patient deaths with a median follow-up of 52 weeks. Figure 2 presents the network diagram. Twelve of twenty-two drugs or drug combinations were directly connected to placebo. Two drug combinations (pirfenidone +N-acetylcysteine, pirfenidone +sildenafil) were connected to the network by pirfenidone, the combination of nintedanib and sildenafil by nintedanib and six (cyclosporine, cyclophosphamide, corticosteroids, corticosteroids+warfarin, corticosteroids+azathioprine, and colchicine) by interferon gamma 1b. The validity of comparisons not directly addressed in clinical trials (ie, cyclophosphamide vs placebo) is dependent on the direct comparisons informing network estimates (ie, interferon gamma 1b vs placebo). These features were similar across the networks. Table 2 contains the summary of findings.

Figure 2

Network diagram for mortality. Each node represents a drug or drug combination that has been tested in trials, the size of the nodes is proportional to the number of patients that have received that drug or drug combination, and the thickness of the edges is proportional to the number of trials.Legend: ERA, endothelin receptor antagonist; NAC, N-acetylcysteine. Triple therapy, corticosteroids+azathioprine+N-acetylcysteine.

Nintedanib (RR 0.69 (0.44 to 1.1)), pirfenidone (RR 0.63 (0.37 to 1.09); direct estimate), and sildenafil (RR 0.44 (0.16 to 1.09)) probably reduce mortality as compared with placebo (all moderate certainty). The combination of pirfenidone +N-acetylcysteine may reduce mortality (RR 0.7 (0.36 to 1.44)) and pamrevlumab may reduce mortality (RR 0.39 (0.1 to 1.45)) as compared with placebo.

Colchicine (RR 2.96 (0.81 to 11.89)) and warfarin (RR 3.85 (1.1 to 16.7)) may increase mortality as compared with placebo, and corticosteroids+azathioprine+N-acetylcysteine (RR 4.76 (0.95 to 50); high certainty) probably increases the risk of death versus placebo.

We did not find any evidence of incoherence.

Change in FVC

Thirty-eight studies reported on FVC, including 8443 patients with IPF. The median follow-up was 48 weeks.

Pamrevlumab probably reduces decline in FVC (per cent predicted) versus placebo (4.3% (0.5 to 8.1); moderate certainty). Pentraxin probably reduces decline in FVC versus placebo (2.74% (1 to 4.83); moderate certainty). Nintedanib probably reduces decline in FVC versus placebo (2.92% (1.51 to 4.14); moderate certainty). Pirfenidone probably reduces decline in FVC versus placebo (2.5% (−0.1 to 5); moderate certainty). The combination of nintedanib +sildenafil versus placebo probably reduces decline in FVC (157.42 mL (−88.35 to 411.12); moderate certainty).

Colchicine (−12.12% (−19.62 to −4.81); moderate certainty), corticosteroids (−14% (−23.56 to −4.46); moderate certainty) and corticosteroids+azathioprine (−9.17% (−20.05 to 1.62); moderate certainty) probably increase the decline of the FVC when compared with placebo.

There was no incoherence in our analysis.

Change in 6-MWD

There were 24 studies included in the 6-MWD with 4353 patients with IPF included. The median follow-up is 48 weeks. Pirfenidone+N-acetylcysteine may improve 6-MWD distance (38.4 (-6.1 to 86.2); low cerainty). There was no evidence of incoherence.

Acute exacerbations and hospitalisations

Thirty-two studies reported on acute exacerbations and hospitalisations, including 8246 patients with IPF with 836 events. The median follow-up was 52 weeks.

Sildenafil probably reduces the risk of acute exacerbations and hospitalisations versus placebo (RR 0.31 (0.05 to 1.4); moderate certainty). Apart from sildenafil, no other medical therapy reduces acute exacerbations and hospitalisations. Nintedanib (RR 0.73 (0.43 to 1.17); high certainty) and pentraxin (RR 0.6 (0.18 to 2.05); moderate certainty) may reduce acute exacerbations and hospitalisations by 2.7% and 3.7%, respectively, but failed to meet a minimally important difference of 5%.

Imatinib (RR 8.15 (0.94 to 248.25); low certainty) and warfarin (RR 2.04 (1.02 to 4); low certainty) may increase the risk of acute exacerbations and hospitalisations. Corticosteroids+azathioprine+N-acetylcysteine (RR 3.7 (1.1 to 14.3); moderate certainty) probably increases the risk of acute exacerbations and hospitalizations. There was no evidence of incoherence.

Adverse events leading to discontinuation

Thirty-seven studies reported on adverse effects, leading to drug discontinuation, including 9033 patients with IPF and 40 studies reported on severe adverse effects, including 9398 patients with IPF. The median follow-up was 52 weeks. Pentraxin (3.17 (0.43 to 93.3); low certainty) may increase adverse events leading to drug discontinuation. Although nintedanib and pirfenidone did not meet our minimally important difference threshold, they both had amongst the highest discontinuation rates.

There was no evidence of incoherence in either network.

Change in DLCO and SAE

Online supplemental appendix 1, p41, presents detailed results for DLCO and SAE. Available treatments probably do not improve DLCO. Corticosteroids+azathioprine+N-acetylcysteine (RR 3.03 (1.4 to 6.25); high certainty) increases the risk of SAE.

Discussion

Main findings

This systematic review and network meta-analysis is the most comprehensive and up to date summary of the evidence for medical treatment in IPF. Our results included 48 RCTs on IPF up until April 2021. Novel investigational treatments such as pamrevlumab, pentraxin and simtuzumab have not been directly compared in trials. We are the first to report the relative effect and safety of these treatments.

In relation to previous findings

We observed similar findings regarding the comparative effectiveness of nintedanib, pirfenidone and sildenafil when compared with other NMAs.2 3 65 66 Distinctions with previous NMAs are that we provide estimates of novel treatments and add new data from the most recent trials to evaluating existing drug treatments. An advantage of our analysis is we use the most up-to-date GRADE guidance for rating the certainty of evidence for NMA. This changes the interpretation of the point estimates from previous studies.

Novel findings

Our review applies state-of-the-art GRADE guidance and addresses a wider range of therapies and includes a greater number of trials than previous reviews. Our review includes 29 trials never before included in other network meta-analyses, including 10 329 participants and 13 new drug nodes

Pamrevlumab is an investigational drug currently evaluated for the treatments of IPF. It is a recombinant human monoclonal antibody against connective tissue growth factor that is a secreted glycoprotein with a presumably important role in fibrogenesis.67 Previously, benefits for pamrevlumab in FVC versus placebo were reported, but no statistically significant mortality was seen.68 Our results support a possible benefit in patients with IPF. However, these findings need to be considered within the investigational nature of this drug and are based on a single trial (PRAISE). The compound is currently in a large phase 3 trial (NCT03955146), which will determine whether this drug will enter the market and become used in clinical practice.

Sildenafil has been proposed as a potential drug therapy for patients with IPF. It has been recently studied in the INSTAGE trial and Behr 2021 as an adjunct with approved antifibrotic therapy.63 69 Our results are similar with Rochwerg et al, who reported moderate certainty evidence of reduction in mortality with sildenafil.2 Both these results are limited in its generalisability, as the majority of the patients included in these studies had advanced lung function impairment. However, some randomised trials show that specific subgroups of patients with IPF may benefit from sildenafil.70 There is growing interest in treating pulmonary vascular disease in patients with IPF. A post hoc analysis showed inhaled treprostinil, a prostacyclin analogue, which has been demonstrated to increase exercise and lung function in patients with IPF.71 Our NMA highlights the potential for sildenafil in the future of IPF, but more study needs to be completed to clearly support its role in managing this disease.

Pentraxin is another investigational drug in IPF. It is a novel recombinant human pentraxin 2 protein that has been shown to have antifibrotic properties and has been demonstrated to be beneficial in bleomycin-associated fibrosis.72 73 It has been studied in two-phase two randomised trials.74 75 We are the first to present pentraxin in a review of the literature for IPF and show relative benefit in reducing decline in FVC. Mortality data were not reported. Further study needs to be done to clarify the role of pentraxins in IPF care and the ongoing phase 3 (NCT04552899) trial will be crucial to better understanding the efficacy of this drug. Currently, our results are limited by the limited number of participants included in the existing trials (van der Blink et al (N=21); Rhagu et al(N=117)).

Implications

The current ATS/ERS/JRS/ALAT Clinical Practice Guidelines conditionally recommend the use of pirfenidone and nintedanib for patients with IPF and provide a conditional recommendation against the use of sildenafil.76 Our review supports the current recommendations in favour of nintedanib and pirfenidone but differ on the recommendation against sildenafil. Since the guidelines, there have been two additional studies with sildenafil in patients with IPF: INSTAGE and Behr 2021.59 69 Further study is needed to elucidate the possible benefits of sildenafil. Future guideline recommendations will likely address pentraxin or pamrevlumab for the treatment of IPF as more randomised trials are completed.

Strength and limitations

The strengths of this review include a comprehensive search strategy, duplicate screening and extraction and a focus on patient-important outcomes. We assessed the certainty of the evidence using the most up-to-date GRADE methodology for NMA. Furthermore, we present the largest and most comprehensive NMA on IPF completed to date.

Our results are limited by possible violations in the assumptions for network meta-analysis, including possible intransitivity (dissimilarity of all comparisons in important factors that may affect the outcome being investigated) and coherence (consistency between direct and indirect estimates). To assess for intransitivity, we assessed the distribution of potential effect modifiers across trials. We considered disease severity as the only convincing potential effect modifier, but the distribution of patient severities was consistent across comparisons. We tested for coherence between direct and indirect estimates by node splitting and did not find evidence of incoherence. Limitations include possible errors in data collection and screening. We identified limited evidence beyond 52 weeks of follow-up. There were also variable methods of IPF diagnosis. Furthermore, our estimates of the effect across outcomes for pirfenidone and nintedanib, which have become standard of care since 2014, may be diluted in our analysis as a result.

Conclusion

We present the largest NMA on IPF medical treatments, focusing on patient important outcomes. We are the first to provide estimates of the comparative efficacy of several novel treatments investigated for IPF. Future studies and guidelines should address the role of novel treatments such as pentraxin and pamrevlumab in patients with IPF and revisit the role of sildenafil.

Data availability statement

Data are available upon reasonable request.

Ethics statements

Patient consent for publication

Ethics approval

This study does not involve human participants.

Acknowledgments

MK discloses research funding for preclinical work from Boehringer Ingelheim and Pieris. He received research funding for clinical projects from Roche. He has received consulting fees from Boehringer Ingelheim, Roche, Horizon, Cipla, Abbvie, Bellerophon, Algernon and CSL Behring. He has received payments or honoraria for lectures, presentations, speaker bruins, manuscript writings or educational events from Novartis, Boehringer Ingelheim and Roche. He has received payments for expert testimony from Roche. He participates on the data safety monitoring board or advisory board for Covance and United Therapeutics. He is the chief editor for the European Respiratory Journal and received a Chief Editor allowance. No other authors have conflicts of interest to disclose. CS discloses research funding from the Canadian Pulmonary Fibrosis Foundation and Boehringer Ingelheim. He has also received payments for presentations and advisory work from Boehringer Ingelheim.

References

Supplementary materials

Footnotes

  • AJ and DZ are joint senior authors.

  • Twitter @tylerpitre6, @faranKhalidPK, @denazera

  • Contributors TP is the main author and the study guarantor. He came up with the study design, developed the methods, recruited the team and led the analytics. He helped write, edit and approve the protocol. He wrote the first draft of the manuscript. JM helped design the methods, she developed the search strategy in conjunction with an academic librarian and trained the data collectors on risk of bias assessment. She helped write, edit and approve the protocol. SC, MZ, RH, MFK, WH and JS screened abstracts, full-text screening and performed data collection. They also performed the risk of bias assessment. All of these tasks were adjudicated and reviewed by TP and JM. LB provided expertise in search strategy and data collection. BG, JC and CS are respirologists who consulted on development of treatment nodes, copy edited and helped review and write the manuscript.MK helped supervise, provide guidance in writing the manuscript and expert advice in idiopathic pulmonary fibrosis. DZ co-supervised the study, consulted in applying risk of bias tools and GRADE, as well as methodology. AJ co-supervised the study. He is the analytic lead and consulting on methodology. All authors reviewed and approved the manuscript.

  • Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.

  • Competing interests MK discloses research funding for preclinical work from Boehringer Ingelheim and Pieris. He received research funding for clinical projects from Roche. He has received consulting fees from Boehringer Ingelheim, Roche, Horizon, Cipla, Abbvie, Bellerophon, Algernon and CSL Behring. He has received payments or honoraria for lectures, presentations, speaker bruins, manuscript writings or educational events from Novartis, Boehringer Ingelheim and Roche. He has received payments for expert testimony from Roche. He participates on the data safety monitoring board or advisory board for Covance and United Therapeutics. He is the chief editor for the European Respiratory Journal and received a Chief Editor allowance. No other authors have conflicts of interest to disclose. CS discloses research funding from the Canadian Pulmonary Fibrosis Foundation and Boehringer Ingelheim. He has also received payments for presentations and advisory work from Boehringer Ingelheim.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.