Nintedanib, Pirfenidone and Pirfenidone Versus Nintedanib: A Systematic Review And Meta-Analysis


 Background: Patients with idiopathic pulmonary fibrosis have a poor overall prognosis. Only nintedanib and pirfenidone have been shown to reduce mortality. Objective: This systematic review and meta-analysis aims to assess the efficacy of nintedanib, pirfenidone, and pirfenidone vs nintedanib on patient important outcomes. Methods: Randomized trials were retrieved from MEDLINE, Cochrane, and EMBASE. The primary outcome was mortality. The secondary outcomes included change in FVC, acute exacerbations and hospitalizations and adverse drug effects leading to discontinuation. We used an inverse variance random effects meta-analysis method to calculate pooled relative risk (RR), standardized mean difference (SMD) and mean difference (MD).Results: A total of 13 studies were included. Both nintedanib [RR 0.63 (0.47,0.85); moderate certainty] and pirfenidone [RR 0.68 (0.47,0.99); moderate certainty] probably reduce all-cause mortality when compared to placebo, but only nintedanib [SMD 0.47 (0.34, 0.60); high certainty] reduces change in FVC. Nintedanib [RR 0.69 (0.48,0.99); moderate certainty],but not pirfenidone probably reduces acute exacerbations or hospitalizations compared to placebo. Compared with placebo, neither nintedanib nor pirfenidone increased risk of drug discontinuation due to adverse effect but there is probably risk of patient drug discontinuation with pirfenidone compared to nintedanib [RR 4.34 (1.72 to 10.98); moderate certainty].Conclusion: Both nintedanib and pirfenidone probably reduce all-cause mortality. Nintedanib is probably more tolerable to pirfenidone in regard to compliance and may be more effective than pirfenidone in reducing mortality rate and in slowing disease progression. Larger head to head randomized trials are needed.


Introduction
Idiopathic pulmonary brosis (IPF) is a form of chronic progressive lung disease of unknown origin. 1,2 The cause is unknown, but likely results from a complex interplay between genetic and environmental factors. The prognosis for IPF patients is unfavourable, with median survival time from diagnosis ranging from 2-4 years. 1 Pharmacologic treatment aims to slow disease progression. 3 Current guidelines on IPF treatment recommend the use of nintedanib and pirfenidone. 4 Both nintedanib and pirfenidone have been shown to reduce the rate of FVC decline and mortality in IPF patients. Several other studies have investigated these rstline treatments. 5 It is unclear whether there is relative superiority of one or the other in clinical practice.
This systematic review and meta-analysis looked at the e cacy of nintedanib, pirfenidone, and pirfenidone vs nintedanib. Patient important outcomes include all-cause mortality, change in forced vital capacity (FVC), acute exacerbations and hospitalizations, as well as adverse effects leading to drug discontinuation.

Methods
This review is registered on Open Science Framework (OSF). We followed the Preferred Reporting Items for Systematic Review and Meta-Analysis (PRISMA) to structure our manuscript.

Search strategy
We conducted a systematic literature review of randomized controlled studies conducted up to April 2021. The databases selected were relevant to the review topic and types of eligible studies. Eligible studies including randomized controlled trials that included treatment with nintedanib or pirfenidone versus placebo or pirfenidone versus nintedanib (vice versa). We developed the electronic search in the bibliographic database MEDLINE in consultation with a professional librarian. We translated the search into EMBASE, Cochrane Library Central/SR, and Clinicaltrials.gov. Supplementary le 1 and 2 have the search strategy.

Eligibility criteria
We included all bibliographic records screened if the population included in the study were adults > = 18 years of age diagnosed with IPF, and the intervention had to include either nintedanib versus placebo, pirfenidone versus placebo or nintedanib versus pirfenidone (vice versa).
The studies included phase 2 or 3 randomized trials (RT). This was chosen to include the highest quality evidence for a research question of this nature.

Data management & selection process
We used COVIDENCE, a systematic review web platform, to screen abstracts. Duplicates were removed prior to uploading using a reference manager. We formed two independent teams to use COVIDENCE to screen titles and abstracts independently and in duplicate for inclusion based on the PICOS design. Title abstracts were screened by 3 independent teams. After screening, full texts were obtained and assessed for eligibility after being uploaded to COVIDENCE by the same three teams. Discrepancies in screening at both stages were resolved through consensus mediation.

Data collection process & data outcomes
We extracted data using a modi ed version of Cochrane's template data collection form for intervention reviews of RCTs.
The primary outcomes included all cause mortality, standardized mean change in FVC, hospitalizations and exacerbations, as well as adverse events leading to discontinuation.

Risk of bias in individual studies
For each included trial, the same three teams used a revision of the Cochrane tool for assessing risk of bias in randomised trials (RoB 2.0) to give a rating to trials. 6 We rated the evidence using the following scale: at i) low risk of bias, ii) some concerns (probably low risk of bias) iii) some concerns (probably high risk of bias) or iv) 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, and including deviations from the registered protocol. Reviewers resolved discrepancies by discussion and, when not possible, with adjudication. See supplementary le 3 for details on our RoB tool.

Data synthesis
We performed an inverse variance meta-analysis using a random effect model, reporting 95% CI. We reported relative risk for our dichotomous outcomes and either mean difference (MD) or standardized mean differences (SMD) for our continuous outcomes. We chose standardized mean differences because FVC was calculated either as change in mL or percent change predicted. We investigated the heterogeneity by the I 2 statistic according to which values of 25%, 50%, and 75% indicate low, moderate, and high heterogeneity levels, respectively. We did not assess publication bias because the number of studies included was less than 10. For studies in which participants were rerandomized, we included in the primary analysis but included a sensitivity analysis in the supplementary material. We used RevMan for all analysis. 7

Con dence in evidence
We assessed the certainty of the evidence with the grading of recommendations assessment, development and evaluation (GRADE), using a minimally contextualized framework. For mortality, 2% reduction was selected as minimally important. We used a standardized mean change of 0.25 as minimally important for our continuous outcomes for mixed mL and percent predicted FVC measurements. For only change in mL, we used a cut off of a 100 mL as minimally important. For acute exacerbations, we used 5% as minimally important and 20% for adverse effects leading to discontinuation. The GRADE approach involves separate grading of quality of evidence for each patient-important outcome followed by determining an overall quality of evidence across outcomes. Two reviewers rated each domain for each comparison separately and resolved discrepancies by consensus. 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, and imprecision.

Study selection
Our search strategy included 2628 records initially and we excluded 1095 from the total for duplication. Therefore, 1533 remained for screening. A total of 50 studies were selected for full text review and 13 randomized controlled trials were included for the systematic review and meta-analysis. See Fig. 1 for the complete PRISMA diagram.

Study characteristics
Thirteen studies reported on outcomes of interest, including randomized 4415 IPF patients. The age and sex of the particiants were similar across all studies, usually older males over the age of sixty. Pulmonary function results were similar across studies. Table 1 has more details on the study characteristics. 8-18

Risk of bias
Most studies were not rated at high risk of bias. One study was at risk of bias due to the randomization process, two due to deviations of the intended interventions, one due to bias in the measurement of the outcome, and one for bias in selection of reported results. Figure 2 has more detail on the risk of bias.  ⨁⨁⨁⨁ HIGH *The risk in the intervention group (and its 95% con dence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CI: Con dence interval; RR: Risk ratio; SMD: Standardised mean difference GRADE Working Group grades of evidence High certainty: We are very con dent that the true effect lies close to that of the estimate of the effect Moderate certainty: We are moderately con dent in the effect estimate: The true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different Low certainty: Our con dence in the effect estimate is limited: The true effect may be substantially different from the estimate of the effect Very low certainty: We have very little con dence in the effect estimate: The true effect is likely to be substantially different from the estimate of effect Explanations a. The upper bound of the con dence intervals is less than a 2% reduction in mortality.
Five studies reported on mortality for pirfenidone. Pooled results of 836 patients show that pirfenidone probably reduces all-cause mortality when compared to placebo [RR 0.68 (0.47,0.99)], I 2 = 0; moderate certainty], using a 2% reduction as minimally important. Forest plot 2 and Table 3 has more details. ⨁⨁⨁⨁ HIGH *The risk in the intervention group (and its 95% con dence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CI: Con dence interval; RR: Risk ratio; SMD: Standardised mean difference GRADE Working Group grades of evidence High certainty: We are very con dent that the true effect lies close to that of the estimate of the effect Moderate certainty: We are moderately con dent in the effect estimate: The true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different Low certainty: Our con dence in the effect estimate is limited: The true effect may be substantially different from the estimate of the effect Very low certainty: We have very little con dence in the effect estimate: The true effect is likely to be substantially different from the estimate of effect Explanations a. The upper bound of the CI is less than a 2% reduction in mortality.
b. The upper bound is greater than the pre-speci ed value of clinical signi cance but the lower bound is much less.
Two studies reported on mortality for pirfenidone versus nintedanib. Pooled results from 140 patients show that there may be no mortality bene t for pirfenidone versus nintedanib, [RR 0.98 (0.14,6.70), I 2 = 0; low certainty, using a 2% reduction in mortality as minimally important. Forest plot 3 and Table 4 has more details.  Table 3 has more details.
Two trials reported on acute exacerbations and all-cause hospitalizations outcome. Pooled results from 69 patients found there may be no bene t for pirfenidone versus nintedanib, [RR 1.56 (0.20 ,12.09), I 2 = 0; low certainty]. Forest plot 9 and Table 4 has more details.

Adverse events leading to drug discontinuation
Five studies reported on adverse effects leading to drug discontinuation for nintedanib. Pooled results from 2409 patients found no risk of discontinuation of nintedanib compared to placebo, [RR 1.23 (0.77, 1.95), I 2 = 76%; high certainty], using a minimally important difference of 20%. Forest plot 10 and summary of ndings  Table 4 has more details.

Main ndings
Our review presents the most up-to-date and highest certainty evidence on the evidence of nintedanib, pirfenidone, and pirfenidone versus nintedanib therapy for IPF patients. We found that both nintedanib and pirfenidone reduce mortality, but nintedanib improved lung function more than pirfenidone. Neither nintedanib or pirfenidone were effective at reducing acute exacerbations or hospitalizations, but nintedanib approached minimally important differences compared to pirfenidone. Compared with placebo, neither patients taking nintedanib or pirfenidone were at risk of drug discontinuation due to adverse effects, but when compared to one another, there was a signi cant risk of patients discontinuation of pirfenidone compared to nintedanib.

Relation to previous ndings
Our ndings of direct comparison showed nintedanib and pirfenidone were both more e cacious versus placebo in reducing mortality, and disease progression. This aligns with several SR's and MA's which have similarly reported the e cacy of nintedanib and pirfenidone in reducing mortality and disease progression in IPF patients. [19][20][21][22] Although in our analysis, FVC did not meet minimal important difference for change in FVC, it was approaching signi cance.
Both medications are currently given a conditional recommendation to use in IPF patients. There has not been a de nitive randomized controlled trial that shows the relative e cacy of one or the other. In terms of safety and adherence, an open label study comparing the safety and tolerability of nintedanib and pirfenidone showed that nintedanib was associated with more treatment associated adverse effects leading to discontinuation. This is contrary to our ndings. However, this study only included patients who already tolerated pirfenidone, which may have biased the results in favour of pirfenidone. 23 Our ndings suggest that nintedanib is more e cacious on a number of fronts. In terms of mortality, there is a greater absolute reduction with nintedanib, as well as improvement of lung function. Drug safety and tolerability are important considerations in patient treatment, and our ndings for adverse effects leading to discontinuation, combined with mortality, improvement of lung function and acute exacerbations and hospitalizations, strongly suggest that nintedanib is more e cacious and tolerable for patients with IPF.
Larger head to head trials should be considered to determine the true point estimate of effects in patient important outcomes.

Strengths and limitations
The strengths of this review include our comprehensive search strategy, data screening and extraction in duplicate, rigorous assessment of risk of bias of trials, and use of the GRADE approach in reporting the certainty in the evidence. Additionally, our review includes studies containing direct, head-to-head trials between nintedanib and pirfenidone, rather than use of purely indirect comparisons seen in past reviews This study has several limitations that should be recognized. There were some differences between studies included in our review, with not all studies measuring all the outcomes, and we were not able to evaluate other important outcomes in our meta-analysis, such as 6-minute walk test, diffusion capacity for carbon monoxide, or quality of life indices. Additionally, there was some variation across studies in treatment duration and follow-up time, preventing inference on long-term outcomes.

Conclusion
This meta-analysis provides the most comprehensive summary to date of the evidence on nintedanib, pirfenidone, and pirfenidone versus nintedanib treatment for IPF patients. We found that nintedanib was probably more e cacious in most patient important outcomes analyzed and likely more tolerable.
Larger randomized trials are needed to con rm this.

Declarations
Ethics approval and consent to participate: Not applicable