Pirfenidone and nintedanib for pulmonary fibrosis in clinical practice: Tolerability and adverse drug reactions

Abbreviations

INTRODUCTION

Idiopathic pulmonary fibrosis (IPF) is the most common form of idiopathic interstitial pneumonia.1 IPF demonstrates a progressive and debilitating disease course with a median survival of 2.5–3.5 years from the time of diagnosis.2 The recent Food and Drug Administration (FDA) approval of nintedanib and pirfenidone has transformed the treatment of IPF, which was previously limited to supportive care and lung transplantation.3 Pirfenidone, a drug demonstrating both anti-inflammatory and antifibrotic properties, has been shown to reduce levels of TNF-α, platelet-derived growth factor (PDGF) and transforming growth factor-β amongst others.4 Nintedanib, an intracellular tyrosine kinase inhibitor, exhibits antifibrotic effects by acting on receptors for PDGF, vascular endothelial growth factor (VEGF) and fibroblast growth factor (FGF).5-7 Randomized clinical trials for nintedanib and pirfenidone both showed a reduction in the rate of decline of lung function with acceptable adverse effect profiles and drug tolerability.8-10

In the two largest clinical trials evaluating pirfenidone in IPF, the CAPACITY and ASCEND trials, the most common adverse drug reactions reported were nausea, diarrhoea, dyspepsia and rash.8, 10, 11 The RECAP trial, an ongoing extension study providing long-term follow-up for patients who participated in the CAPACITY trial, showed similar adverse event rates with nausea occurring most frequently.12 In the INPULSIS 1 and 2 trials, the largest studies to date assessing the efficacy of nintedanib for treatment of IPF, gastrointestinal-related adverse reactions (i.e. diarrhoea) were most frequently reported.9 The INPULSIS-ON study, an ongoing open-label extension study for nintedanib, has reported preliminary results suggesting a similar safety and drug tolerability profile as INPULSIS 1 and 2.13

In clinical practice, many patients who receive antifibrotic therapy have significant medical co-morbidities or baseline characteristics that would have excluded them from clinical trial participation. For example, subjects with severe abnormalities on pulmonary function test (PFT) (forced vital capacity (FVC) < 50% predicted and diffusion capacity of carbon monoxide (DLCO) < 30% predicted) were excluded from the previously mentioned clinical trials. We hypothesized that the drug tolerability and adverse events of pirfenidone and nintedanib therapy would differ from prior studies in a more heterogeneous patient population including subjects with advanced disease, a high prevalence of co-morbid medical conditions and the inclusion of subjects with non-IPF pulmonary fibrosis in addition to IPF.

METHODS

RESULTS

A total of 186 subjects from the 249 reviewed met inclusion criteria for the study. Subjects were excluded for the following reasons: 49 subjects opted out of antifibrotic therapy after the prior authorization paper work was initiated, 5 subjects were unable to obtain insurance approval for drug, 7 subjects died or transitioned to home hospice prior to drug initiation and 2 subjects underwent lung transplantation prior to starting drug. Of the 186 subjects included, 129 were prescribed pirfenidone and 57 received nintedanib. The mean observation period for subjects on pirfenidone was 52 ± 17 weeks, and for nintedanib was 41 ± 15 weeks.

Baseline characteristics for subjects in our cohort are listed in Table 1. Corresponding values from pertinent clinical trials are reported when available. One hundred fifty-two of the 186 subjects (81.7%) were prescribed antifibrotic therapy for IPF. Remaining subjects had fibrotic ILD without typical features of UIP on HRCT or surgical lung biopsy (in these subjects, pulmonary fibrosis was believed to be the driver of disease by the subject's primary pulmonologist). Overall, the study population was comprised of individuals with advanced pulmonary fibrosis as evidenced by the large proportion requiring home oxygen therapy (117 of 186 subjects or 63%) and a mean DLCO % predicted of 36 ± 14. Subjects in our study had a lower mean DLCO and FVC % predicted than subjects from corresponding clinical trials (Table 2). Medical co-morbidities with the highest prevalence in our cohort were coronary artery disease (50 of 186 subjects or 27%) and gastroesophageal reflux disease (GERD) (95 of 186 subjects or 51%). Evaluation of HRCT revealed either a definite or a possible UIP pattern in the majority of subjects (Table 2).

Adverse events occurring with the highest frequency for subjects on pirfenidone were nausea (26.4%), rash/photosensitivity (14.7%) and dyspepsia/GERD (12.4%). Anorexia, vomiting, diarrhoea and rash/photosensitivity occurred less frequently compared with pooled results from the CAPACITY and ASCEND trials (Table 3). On nintedanib therapy, diarrhoea (52.6%) and nausea (29.8%) were reported most often. Adverse reactions on nintedanib occurred with similar frequencies when compared with pooled data from the INPULSIS 1 and 2 trials.

The outcomes for subjects experiencing any adverse reaction are shown in Figure 1. 66.7% of subjects tolerated pirfenidone without dose reduction nor drug discontinuation. In comparison, subjects treated with nintedanib tolerated therapy in 52.6% of cases. A dose reduction was necessary in 12.4% of subjects prescribed pirfenidone, and in 21.1% of subjects on nintedanib. None of these differences reached statistical significance.

Drug discontinuation rates as a result of an adverse event were similar for pirfenidone and nintedanib (20.9% vs 26.3%, respectively). The rates of drug discontinuation did not significantly differ when compared with corresponding clinical trials (Table 3). Pirfenidone discontinuation resulted most often from a skin-associated adverse reaction (10 of 27 subjects) or a gastrointestinal-associated adverse reaction (10 of 27 subjects). In contrast, nintedanib was discontinued as a result of a gastrointestinal-related adverse event in the majority of cases (Fig. 2).

One subject on nintedanib had a severe bleeding event from a duodenal ulcer requiring hospitalization, multiple blood transfusions and an upper endoscopy with placement of two endoclips. Of note, this patient was 76 years old and on dual antiplatelet therapy with aspirin and clopidogrel at the time of the upper gastrointestinal bleed.

No patients on nintedanib had an acute myocardial infarction during the study. One subject on pirfenidone, a 71-year-old male with a known history of coronary artery disease, had a non-ST segment elevation myocardial infarction requiring placement of multiple stents. His hospitalization was also complicated by new-onset atrial fibrillation.

In our study, 16.3% of subjects on pirfenidone and 10.5% of subjects on nintedanib had at least one acute exacerbation of pulmonary fibrosis (Table 3). This trend towards fewer exacerbations in the nintedanib group as compared with pirfenidone did not reach statistical significance (P = 0.37). Patients receiving nintedanib had a trend towards prolonged time to first exacerbation compared with pirfenidone (mean duration: 294 days vs 247 days), but this also was non-significant (P = 0.72).

We performed a multivariate logistic regression analysis to evaluate for clinical characteristics associated with drug discontinuation. A history of congestive heart failure (OR: 4.54, 95% CI: 1.41–14.55, P = 0.01) and an age ≥70 years old (OR: 2.21, 95% CI: 1.04–4.69, P = 0.04) were the only characteristics associated with antifibrotic drug discontinuation. Of note, clinical characteristics associated with disease severity (FVC % predicted, DLCO % predicted and home oxygen use) had no association with drug discontinuation. Additionally, we compared subjects with non-IPF pulmonary fibrosis to subjects with IPF using propensity score matching to evaluate the association of ILD diagnosis on drug tolerability. After statistical matching, we found that subjects with non-IPF pulmonary fibrosis had similar drug discontinuation and adverse reaction rates to those with IPF (Table S2).

Subjects were compared for differences in drug tolerability with varying lung function. Using cut-off values of 50% predicted for FVC and 30% predicted for DLCO (inclusion criteria cut-off values in clinical trials), we found no statistically significant differences in the rates of common adverse reactions or drug discontinuation for nintedanib or pirfenidone (Fig. 3).

Fifteen subjects in the study crossed over and were prescribed the alternative antifibrotic drug after demonstrating intolerability to the first therapy (Table S3). Eleven of these subjects switched to nintedanib, while the other four transitioned to pirfenidone. Nine of the subjects (60%) tolerated the alternative therapy, three (20%) developed an adverse reaction necessitating a dose reduction and three (20%) discontinued the alternative drug as well.

DISCUSSION

Our study evaluated the drug tolerability and adverse effect profiles of nintedanib and pirfenidone for pulmonary fibrosis at a large ILD referral centre. Despite our subjects having greater baseline respiratory impairment and significant co-morbid medical conditions, adverse events and drug discontinuation occurred with a frequency similar to clinical trial data. In fact, in our study, subjects receiving pirfenidone reported anorexia, vomiting, diarrhoea and rash less frequently than subjects from the ASCEND and CAPACITY trials.

Subjects in our study who received nintedanib had poorer baseline lung function (lower mean DLCO and FVC % predicted) compared with subjects in the INPULSIS 1 and 2 trials.9 In addition, the majority of our subjects were on home supplemental oxygen suggesting advanced pulmonary fibrosis. Despite these traits, subjects from our cohort had comparable drug tolerability and adverse event rates to those from the INPULSIS 1 and 2 trials. Our findings suggest that nintedanib may be prescribed to patients with advanced pulmonary fibrosis without greater risk of drug intolerability.

Pirfenidone, which only gained FDA approval in 2014, has been approved for treatment of IPF in Japan since 2008 and in Europe since 2011. In subsequent years, several publications have reported on the tolerability of pirfenidone in both Europe and Japan. A study by Ogura et al. reported on the Japanese post-marketing surveillance for 1371 IPF patients treated with pirfenidone. The authors concluded that pirfenidone had a comparable safety profile to previous clinical trials in Japan with an adverse event drug discontinuation rate similar across all stages of IPF severity (22.9–26.7%).14 Conversely, a recent retrospective observational study utilizing surveys evaluated the clinical experience of pirfenidone in 502 patients from Japan showed a higher 1-year drug discontinuation rate of 37.1%.15 Multiple, small, single-centre studies in Europe have reported drug discontinuation rates of pirfenidone ranging from 13% to 19%.16-24 The discontinuation rate of pirfenidone in our study was 20.9%, which is within the range of previously published results.

We report in our study at least one acute exacerbation of pulmonary fibrosis in 16.3% of patients on pirfenidone and 10.5% of patients on nintedanib. The 1-year incidence of IPF exacerbations has varied widely in the literature (3.6–14.7%).9, 25, 26 The incidence of IPF exacerbations in our studies was higher than that reported in most previous studies. This could be a consequence of our population being compromised of subjects with more advanced disease comparatively and our inclusion of subjects with non-IPF pulmonary fibrosis.

One limitation of our study is that information regarding the primary and secondary outcomes was derived from review of the EMR and thus depended on both subjects to actively report adverse reactions to their healthcare providers and for providers to document events in the medical record. As a result, the true incidence of adverse reactions may have been underestimated in our cohort. In contrast, our drug discontinuation rates should not be susceptible to this issue as these events were confirmed by correspondences from the outpatient specialty pharmacies that dispensed nintedanib and pirfenidone. Second limitation was the single-arm design of the study that lacked a control or placebo group. Lastly, subject's diagnosis of IPF was determined by their primary pulmonologist during routine clinical care. This could have led to a confirmation bias with respect to a clinical diagnosis of IPF in some patients, as insurance companies in the United States currently require this diagnosis for coverage of pirfenidone and nintedanib.

In conclusion, our study evaluating the use of nintedanib and pirfenidone at a large ILD referral centre determined that both drugs had an adverse event profiles and drug tolerability rates comparable with previously published clinical trials. This held true despite our cohort having a greater degree of respiratory impairment, a high prevalence of co-morbid medical conditions and the inclusion of subjects with non-IPF pulmonary fibrosis in addition to IPF. While it is promising that each of these agents appear safe to use in a broader cohort of patients than previously included in large randomized controlled trials, future prospective studies are needed to elucidate the efficacy of treatment with expanded prescribing practices.