To the Editor:
Idiopathic pulmonary fibrosis (IPF) is a deadly disease that typically develops after age 60 (1). Given the age of onset, patients with IPF are at risk for other aging-related conditions such as heart disease, pneumonia, and cancer (2). These competing causes of death may introduce bias when utilizing all-cause mortality to conduct causal inference or as an anchor to develop prognostic biomarkers and clinical severity models. To increase confidence that clinical and molecular predictors of death reflect IPF-relevant pathobiology, a better understanding of competing causes of death is needed.
The natural history of IPF is characterized by a progressive decline in FVC (3). Change in FVC remains the preferred outcome measure in IPF clinical trials (4), and a categorical decline in FVC of ≥5% is a common measure of IPF progression (5, 6). Accordingly, accounting for FVC trajectory before death could inform the probability of a death being related to IPF. In this study, we report attributed causes of death for participants in the PROFILE (Prospective Observation of Fibrosis in the Lung Clinical Endpoints) study, a United Kingdom–based multicenter prospective cohort study of IPF (7). Leveraging serially acquired FVC measures, we identified patients with physiological progression before death, hypothesizing that antecedent progression would vary according to attributed cause of death.
All PROFILE participants who died within 60 months of enrollment (n = 383) were eligible for inclusion. We excluded those without baseline FVC measure (n = 6), follow-up FVC measure within 24 months of death (n = 104), and unknown cause of death (n = 32). Change in FVC was calculated using the final two FVC measures before death, and progression was defined as ≥5% relative FVC decline between these measures. Absolute change in measured FVC (in milliliters) between measures was also calculated. Attributed cause of death was determined by a review of death certificates, or by a discussion with the primary care physician when the death certificate could not be obtained, and was categorized as due to IPF or other respiratory, malignancy, or cardiac causes. Death due to isolated pulmonary vascular disease (n = 7) was presumed to be secondary to IPF unless ischemic heart disease was also noted. A chi-square test was performed to compare expected and observed proportions of patients with antecedent progression across attributed causes of death.
Two hundred forty-one participants met the inclusion criteria and were similar to those who were excluded (Table 1). Of those who were included, 179 (74.3%) displayed an antecedent FVC decline of ≥5%. The most common attributed cause of death was IPF (55.6%), followed by other respiratory causes (26.6%), malignancy (9.5%), and cardiac causes (8.3%) (Figure 1A). When stratifying by attributed cause of death, antecedent progression was observed in 84.3% (n = 113) of deaths attributed to IPF , 67% (n = 43) of deaths attributed to other pulmonary causes, 52% (n = 12) of deaths attributed to malignancy and 55% (n = 11) of deaths attributed to malignancy (P < 0.001) (Figure 1B). In descriptive analysis, relative and absolute change in FVC also varied across attributed causes of death (Table 1).
Table 1.
Baseline Characteristics and Outcomes for Included and Excluded Cases
| Characteristic | Included Patients (n = 241) | Included Cases, Attributed Cause of Death |
Excluded Patients (n = 142) | |||
|---|---|---|---|---|---|---|
| IPF (n = 134) | Other Respiratory (n = 64) | Malignancy (n = 20) | Cardiac (n = 20) | |||
| Demographics | ||||||
| Age, mean (SD) | 70.1 (8.1) | 69.3 (8.1) | 72.0 (8.1) | 69.8 (9.1) | 72.4 (7.1) | 72.9 (8.3) |
| Sex, n (%) | 193 (80.1) | 110 (82.1) | 42 (79.3) | 18 (78.3) | 17 (85.0) | 115 (81.6) |
| Ever smoker, n (%) | 170 (70.8) | 94 (70.7) | 41 (77.4) | 17 (73.9) | 12 (60.0) | 98 (69.5) |
| Comorbidities, n (%) | ||||||
| Hypertension | 133 (57.8) | 68 (53.1) | 38 (63.3) | 12 (54.6) | 15 (75.0) | 59 (45.4) |
| Diabetes mellitus | 39 (17.0) | 22 (17.2) | 8 (13.3) | 6 (27.3) | 3 (15.0) | 23 (17.7) |
| Cancer (any type) | 16 (7.0) | 4 (3.1) | 7 (11.7) | 4 (18.2) | 1 (5.0) | 6 (4.6) |
| Lung function | ||||||
| Baseline FVC, % predicted, mean (SD) | 73.0 (16.8) | 71.0 (17.2) | 73.4 (17.9) | 78.2 (10.9) | 78.6 (17.6) | 71.9 (17.9) |
| Baseline Dl CO, % predicted, mean (SD) | 38.6 (11.1) | 37.2 (10.6) | 40.1 (10.9) | 41.5 (10.3) | 39.5 (13.2) | 36.9 (11.6) |
| Absolute change in FVC (ml), median (IQR) | −310 ml (−550 ml, −120 ml) | −400 ml (−600 ml, −190 ml) | −200 ml (−405 ml, −50 ml) | −170 ml (−490 ml, 90 ml) | −140 ml (−425 ml, 185 ml) | NA |
| Relative change in FVC, %, median (IQR) | −13.6% (−22.9%, −3.8%) | −18.0% (−25.3%, −8.5%) | −8.8% (−18.9%, −1.9%) | −5.4% (−17.7%, 2.7%) | −6.7% (−17.9%, 6.3%) | NA |
| Months between last two PFTs, median (IQR) | 12.5 (8.7, 18.9) | 13.1 (9.5, 20.4) | 11.8 (7.3, 16.0) | 12.9 (8.4, 19.6) | 11.5 (8.1, 14.1) | NA |
| Months from last PFT to death, median (IQR) | 9.0 (4.5, 14.2) | 9.8 (4.5, 14.7) | 8.6 (4.1, 15.0) | 9.3 (4.2, 11.8) | 8.0 (5.4, 10.4) | NA |
Definition of abbreviations: IPF = idiopathic pulmonary fibrosis; IQR = interquartile range; NA = not applicable; PFT = pulmonary function test.
Figure 1.
(A) Bar graphs displaying the attributed causes of death in PROFILE (Prospective Observation of Fibrosis in the Lung Clinical Endpoints) and (B) proportion of patients experiencing IPF progression in the 24 months before death across each attributed cause of death. IPF = idiopathic pulmonary fibrosis.
In this study, we found that 74% of eligible patients who died during the PROFILE study experienced antecedent IPF progression. The proportion of patients with antecedent progression varied significantly according to attributed cause of death, with large majorities of those with death attributed to IPF and other respiratory causes experiencing antecedent FVC decline. Those with death attributed to IPF and other respiratory causes also had higher rates of FVC decline than those with deaths attributed to nonpulmonary conditions. Although the true cause of death remains difficult to ascertain in IPF, we found that those with deaths attributed to IPF and other respiratory conditions were more likely to display a classic IPF natural history than those with deaths attributed to other conditions. These findings increase confidence that most respiratory deaths likely reflect causal biological pathways relevant to IPF.
Our findings build on a prior Finnish study showing a similarly high proportion of competing causes of death in IPF (2). Given the protocolized nature of the PROFILE study, which obtained serial lung function measures through 36 months, our study was also able to assess disease trajectory before death. Beyond increasing confidence that respiratory-related deaths are related to IPF-relevant pathways, our findings also highlight the potential for competing causes of death to introduce bias when modeling all-cause mortality. This can occur because of confounding, whereby a competing cause of death (e.g., heart failure) also influences the predictor of interest (e.g., lung function) (8). Bias can also result when a competing cause of death (e.g., cancer) mediates the relationship between an exposure (e.g., cigarette smoking) and death, producing a valid exposure-outcome association but through biological pathways potentially unrelated to the IPF (8).
Although new drugs that improve all-cause mortality would be welcome in IPF, our findings underscore the potential importance of informative missingness in IPF clinical trials. Several imputation methods have been pursued when the FVC was missing because of death, including the assumption of a zero value (9, 10). If deaths unrelated to IPF are differential by treatment group, FVC replacement using more extreme values (e.g., zero) could introduce more bias when estimating treatment effect than less extreme values. This supports a conservative approach to FVC imputation in IPF clinical trials and rigorous sensitivity analyses of informative missingness to understand the impact of FVC replacement due to death and ensure consistent treatment effect.
Our study has several limitations. First, we relied on externally reported causes of death to conduct our analysis, which is prone to information bias. We attempted to mitigate this bias by focusing on cases with a clearly documented cause of death and secondary review of national health records when necessary. Next, despite the protocolized nature of PROFILE, 27% of patients who died during the study did not undergo follow-up FVC before death, leaving it unclear whether our findings would remain consistent in this subset. Finally, the optimal observation period for monitoring for progression remains unclear.
In summary, this study sheds important light on competing causes of death in IPF and highlights the challenges in identifying prognostic biomarkers and the confidence with which we can assume that specific causes of death reflect IPF-relevant pathobiology.
Footnotes
Supported by GlaxoSmithKline R&D grant R01HL169166 and the Medical Research Council (to the PROFILE study) grant R01HL166290 (to J.M.O.).
Author Contributions: J.M.O. and P.L.M. designed and supervised the study. T.M.M., R.G.J., S.R.J., and P.L.M. recruited patients. L.Y. and P.L.M. reviewed death records. J.M.O. and P.L.M. performed the analysis. A.T., T.M.M., R.G.J., S.R.J., I.S., and P.LM. interpreted the results. A.T., T.M.M., J.M.O., and P.L.M. wrote the manuscript, with input from all authors. All authors approved the final version of the manuscript.
Originally Published in Press as DOI: 10.1164/rccm.202403-0595RL on July 16, 2024
Author disclosures are available with the text of this article at www.atsjournals.org.
References
- 1. Raghu G, Remy-Jardin M, Richeldi L, Thomson CC, Inoue Y, Johkoh T, et al. Idiopathic pulmonary fibrosis (an update) and progressive pulmonary fibrosis in adults: an official ATS/ERS/JRS/ALAT clinical practice guideline. Am J Respir Crit Care Med . 2022;205:e18–e47. doi: 10.1164/rccm.202202-0399ST. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Kärkkäinen M, Nurmi H, Kettunen HP, Selander T, Purokivi M, Kaarteenaho R. Underlying and immediate causes of death in patients with idiopathic pulmonary fibrosis. BMC Pulm Med . 2018;18:69. doi: 10.1186/s12890-018-0642-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Ley B, Collard HR, King TE., Jr Clinical course and prediction of survival in idiopathic pulmonary fibrosis. Am J Respir Crit Care Med . 2011;183:431–440. doi: 10.1164/rccm.201006-0894CI. [DOI] [PubMed] [Google Scholar]
- 4. Raghu G, Ghazipura M, Fleming TR, Aronson KI, Behr J, Brown KK, et al. Meaningful endpoints for idiopathic pulmonary fibrosis (IPF) clinical trials: emphasis on ‘feels, functions, survives.’ Report of a collaborative discussion in a symposium with direct engagement from representatives of patients, investigators, the National Institutes of Health, a patient advocacy organization, and a regulatory agency. Am J Respir Crit Care Med . 2024;209:647–669. doi: 10.1164/rccm.202312-2213SO. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Zappala CJ, Latsi PI, Nicholson AG, Colby TV, Cramer D, Renzoni EA, et al. Marginal decline in forced vital capacity is associated with a poor outcome in idiopathic pulmonary fibrosis. Eur Respir J . 2010;35:830–836. doi: 10.1183/09031936.00155108. [DOI] [PubMed] [Google Scholar]
- 6. Richeldi L, Du Bois RM, Raghu G, Azuma A, Brown KK, Costabel U, et al. Investigators IT Efficacy and safety of nintedanib in idiopathic pulmonary fibrosis. N Engl J Med . 2014;370:2071–2082. doi: 10.1056/NEJMoa1402584. [DOI] [PubMed] [Google Scholar]
- 7. Maher TM. PROFILEing idiopathic pulmonary fibrosis: rethinking biomarker discovery. Eur Respir Rev . 2013;22:148–152. doi: 10.1183/09059180.00000913. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Lederer DJ, Bell SC, Branson RD, Chalmers JD, Marshall R, Maslove DM, et al. Control of confounding and reporting of results in causal inference studies. Guidance for authors from editors of respiratory, sleep, and critical care journals. Ann Am Thorac Soc . 2019;16:22–28. doi: 10.1513/AnnalsATS.201808-564PS. [DOI] [PubMed] [Google Scholar]
- 9. Thabut G, Crestani B, Porcher R, Richeldi L. Missing data in IPF trials: do not let methodological issues undermine a major therapeutic breakthrough. Eur Respir J . 2015;46:607–614. doi: 10.1183/13993003.02204-2014. [DOI] [PubMed] [Google Scholar]
- 10. Lederer DJ, Bradford WZ, Fagan EA, Glaspole I, Glassberg MK, Glasscock KF, et al. Sensitivity analyses of the change in FVC in a Phase 3 trial of pirfenidone for idiopathic pulmonary fibrosis. Chest . 2015;148:196–201. doi: 10.1378/chest.14-2817. [DOI] [PMC free article] [PubMed] [Google Scholar]