Impact of surgical lung biopsy on lung function and survival in patients with idiopathic pulmonary fibrosis in a multi‐centre registry cohort

Establishing an accurate and timely diagnosis of idiopathic pulmonary fibrosis (IPF) is essential for appropriate management and prognostication. In some cases, surgical lung biopsy (SLB) is performed but carries non‐negligible risk. The objective of this retrospective study was to determine if SLB is associated with accelerated lung function decline in patients with IPF using the Canadian Registry for Pulmonary Fibrosis.


INTRODUCTION
Idiopathic pulmonary fibrosis (IPF) is a progressive fibrosing interstitial lung disease (ILD) of complex aetiology. 1Establishing an accurate and timely diagnosis is essential for appropriate management and prognostication.The antifibrotic drugs nintedanib and pirfenidone both reduce the rate of lung function loss over time, 2,3 but there remains no cure.Current clinical practice guidelines recommend an IPF diagnosis after exclusion of known causes and with compatible chest imaging 1,4 ; however, histopathological sampling historically held greater priority and was performed more frequently.While surgical lung biopsy (SLB) can provide contributory histopathologic data supporting diagnosis, it is not without risk.A United Kingdom-based study reported the in-hospital, 30-day and 90-day mortality to be 1.7%, 2.4% and 3.9%. 5In a Canadian study, SLB for ILD carried an estimated 30-day overall mortality of 7.1% with clearly increased risk for non-elective versus elective procedures (20.2% vs. 1.9%). 6Data from a tertiary referral centre in the United States reiterate the differential risk between elective and non-elective procedures.There were no deaths at 1 month in the group undergoing elective procedures, although 8% experienced a complication, and 4% a severe complication.Of patients hospitalized prior to their procedure, the 1-month mortality rate was 10%. 7hile peri-operative mortality risk is well established, there are limited data informing the impact of SLB on lung function in patients with IPF.It is unknown whether SLB results in a single decrement in lung function related to excision of functional lung parenchyma, accelerates decline or has no effect on lung function.A retrospective review conducted on patients with ILD reported no deleterious effect on lung function whether the biopsy was via open or minimally invasive technique. 8However, this study included patients with diagnoses of vasculitis, autoimmune lung disease, and sarcoidosis, limiting generalizability to IPF or other progressive ILDs, and predates the modern diagnostic and treatment guidelines for IPF. 9 A more recent single centre retrospective study reported a mean annual decline of 363 mL (±764) or approximately 5% in forced vital capacity (FVC). 10However, there were no comparisons to non-biopsied patients, or FVC comparisons pre and post procedure.
The objective of this retrospective study was to characterize lung function decline in patients with IPF that underwent SLB compared to those who did not undergo SLB, using longitudinal data from a multi-centre registry. 11We hypothesized that SLB would result in accelerated lung function decline.We also compared the pre-operative and postoperative rates of FVC% and diffusion capacity of the lung (DLCO%) decline within the biopsied patients.

Study population
The Canadian Registry for Pulmonary Fibrosis (CARE-PF) is a multi-centre ILD registry collecting prospective clinical cohort data at eight ILD tertiary care centres across Canada.Diagnoses are established according to contemporaneous clinical practice guidelines where available, and multidisciplinary discussion (MDD).IPF patients enrolled in the registry between January 2015 and May 2019 were included for this retrospective analysis.Patients were excluded if they were missing FVC data, key demographics or censoring dates.

Study design
For the primary analysis, two groups were created.For the non-SLB group, baseline was set as the first ILD clinic visit

SUMMARY AT A GLANCE
The diagnostic evaluation for suspected idiopathic pulmonary fibrosis (IPF) may include surgical lung biopsy to obtain histopathology.In our study, we found no evidence of accelerated lung function decline or increased risk of death or lung transplantation in patients with IPF who underwent surgical lung biopsy.
date, and all patients with ≥2 FVC measurements ≥3 months apart, up to 24 months were included.The SLB group was established similarly, but with baseline set as their date of SLB.The period of 24 months was chosen to ensure sufficient measurements for analysis, and because progression would be expected to manifest in both groups over 24 months, including progression attributable to SLB.The second analysis compared pre-post-SLB rates of lung function decline, with patients requiring ≥2 FVC measurements ≥3 months apart from 1-year pre-to 1-year post-SLB.Patients who underwent SLB could be included in both analyses.

Baseline characteristics
Clinical data collected include baseline age, patient-reported sex, smoking status, race, antifibrotic treatment, absolute lung function (FVC and DLCO) and vital status including dates of death and transplant.Once patients receive transplant there is no further prospective data collected in the registry and transplant date equates to their censor date.Absolute FVC and DLCO were standardized to Global Lung Initiative (GLI) reference ranges to provide FVC% and DLCO% predicted values, including correcting DLCO for ambient barometric pressure at each of the study site locations in Canada. 12,13reatment was defined as use of nintedanib or pirfenidone for ≥3 months consecutively during the period in which lung function data were obtained for analysis.

Statistical analysis
Descriptive statistics were used to analyse baseline characteristics.Unadjusted and adjusted linear mixed effects models compared decline in FVC% over time between groups.Transplant free survival was assessed using Kaplan-Meier plots, and through Cox proportion hazards models with transplant as a competing risk.Models were adjusted for baseline age, sex, smoking history, antifibrotic treatment, FVC% and DLCO%.Assumptions were met in the Cox models.

Study population and characteristics
Of 3267 patients in the registry, 808 had IPF.There were 113 patients who underwent SLB and after exclusion, 81 who met criteria for the primary analysis.Of 695 patients who did not undergo SLB, 468 met criteria for the primary analysis (Figure 1).Baseline characteristics are presented in Table 1.For the SLB group, mean baseline age was 64 (±7) years and the non-SLB group was older with mean age of 71 (±9) years, with the majority being male.For the SLB group mean FVC% was 73% (±15) and for the non-SLB group it was slightly higher at 79% (±19).Groups appeared similar with respect to IPF-specific treatment.There was one patient in the SLB group that was on antifibrotic treatment prior to SLB.Only 15 patients met inclusion criteria for the pre-post-SLB group analysis (Figure 1).The mean baseline age was 63 (±6) years, most were male, with a mean FVC% of 71% (±14).None of the pre-post-SLB patients were on antifibrotic treatment prior to SLB.

Change in lung function
For patients who underwent SLB, mean annual absolute FVC decline was 120 mL (±30), while it was 109 mL F I G U R E 1 Flow diagram for creation of the three groups.The SLB and non-SLB groups were directly compared for decline in lung function over the 24-month period.The pre-post-SLB group compared decline in lung function over 12 months pre-SLB and 12 months post-SLB.FVC, forced vital capacity; ILD, interstitial lung disease; IPF, idiopathic pulmonary fibrosis; n, number; SLB, surgical lung biopsy.
(±12) in the non-SLB group, with no difference between groups in unadjusted ( p = 0.74) or adjusted models (p = 0.75) (Figure 2A).For patients who underwent SLB, mean annual FVC% decline was 2.0% (±0.8) in the unadjusted model and 2.1% (±0.8) in the adjusted model.Mean annual decline in FVC% for the non-SLB group was 2.4% (±0.3) in both unadjusted and adjusted models.Again, there was no difference between groups in FVC%  F I G U R E 2 Adjusted mean annual decline for SLB and non-SLB groups for (A) FVC (L) absolute, (B) FVC% and (C) DLCO% demonstrating no significant difference between groups.DLCO%, diffusion capacity of the lung percent predicted; FVC, forced vital capacity; FVC%, forced vital capacity percent predicted; L, litres; SLB, surgical lung biopsy.
decline in unadjusted (p = 0.67) or adjusted models (p = 0.73) (Figure 2B).For patients who underwent SLB, the mean annual decline in DLCO% was 4.5% (±0.9) in the unadjusted model and 4.6% (±0.9) in the adjusted model.Mean annual decline in DLCO% for the non-SLB group was 4.5% (±0.4) in both unadjusted and adjusted models.There was no difference in mean annual decline of DLCO% in either unadjusted (p = 1.0) or adjusted models ( p = 0.96) (Figure 2C).The proportion of patients with an absolute FVC% decline ≥10% over 24 months was 17% in the SLB group versus 24% in the non-SLB group, similar between groups (p = 0.20).The proportion of patients with an absolute DLCO% decline of ≥15% over 24 months was 30% in the SLB group versus 26% in the non-SLB group, which was also similar ( p = 0.48).

Transplant-free survival
All 113 SLB patients were included in the survival analysis.There was no difference in time to death or lung transplant in the SLB versus the non-SLB group ( p = 0.09) (Figure 5).SLB patients had similar risk of death or transplant compared to non-SLB patients in unadjusted (hazard ratio [HR] = 0.68, 95%CI 0.45-1.03;p = 0.07) and adjusted models (HR = 0.94, 95%CI 0.60-1.47;p = 0.79).Median survival for the SLB group was 11.6 years (interquartile range [IQR] 4.4 to undefined) while it was 7.1 years (IQR 3.6 to undefined) for the non-SLB group.
In the SLB group, there were four transplants and three deaths recorded within 24 months of SLB; one death was attributed to an acute exacerbation of acute exacerbation of idiopathic pulmonary fibrosis (AE-IPF), one to lung cancer, and one without identified cause.In the non-SLB group, there were 22 transplants and 7 deaths recorded within the 24 months of follow-up.No deaths occurred in the prepost-SLB group in the 1-year post-SLB and there was one transplant in the 1-year post-SLB.

DISCUSSION
In this multi-centre registry-based study, we found that SLB was not associated with accelerated lung function decline in patients with IPF compared to lung function decline in those not undergoing SLB.Lung function decline appeared attenuated in the pre-post-SLB comparison; however, findings are limited by the small sample size.Additionally, we found that SLB was not associated with increased risk of death or lung transplantation with no deaths occurring within 1 month as opposed to 2 (1.7%) in the prior study. 10These findings contradict our original hypothesis and the most recent ILD specialized single-centre retrospective study that demonstrated significant lung function decline after biopsy. 10It was presumed that SLB may result in accelerated lung function decline due to the removal of potentially functional lung or from acute exacerbations due to the process of mechanical ventilation and surgical manipulation of lung tissue.Evolving surgical technique and ventilation measures in addition to proper selection of patients for SLB may contribute to our overall findings.Important distinctions from previous data include our inclusion of a control population showing similar FVC decline in the non-SLB group, consistent with the natural history of IPF, and requirement for multiple PFTs.
An accurate diagnosis of IPF remains essential, with international multi-society guidelines highlighting the importance of clinical, radiologic, and pathologic findings considered in the context of MDD. 1,4The recognition of a probable usual interstitial pneumonia (UIP) pattern on high-resolution computed tomography (HRCT) of the chest along with clinical features of sex, age, extent of fibrosis has helped increase diagnostic confidence in IPF without need for invasive testing and lung biopsy. 1,14,15With the introduction of diagnostic likelihood bands, there may be less need for SLB. 16In a study that had respiratory physicians evaluate ILD cases with HRCT chest imaging and assign confidence to their IPF diagnoses, SLB was requested in 8.1%, 29.6%, and 48.4% of definite, provisional high-confidence, and provisional-low confidence diagnoses of IPF. 17 Antifibrotic therapy was prescribed in 63% of the provisional high-confidence IPF diagnoses without requesting SLB.There was no significant mortality difference for those with a definite IPF diagnosis and provisional high-confidence IPF diagnosis, suggesting that a confidence ≥70% favouring an IPF diagnosis appears satisfactory for clinicians to make treatment decisions without need for further invasive testing.
Alternative sampling techniques and approaches to IPF diagnosis are available.Transbronchial biopsy, while less invasive than SLB, has low diagnostic accuracy for fibrotic ILD and introduces risk including pneumothorax. 1 It is not recommended for routine clinical use in the most recent IPF guidelines. 43][24] Endobronchial optical coherence tomography is a promising minimally invasive technique for microscopic diagnosis of ILD, with recent data demonstrating high sensitivity and specificity for histopathologic UIP and clinical diagnosis of IPF. 25 Research on serum biomarkers, including matrix metalloproteinase 7 and surfactant protein D, are ongoing, with likely key roles in distinguishing IPF from other ILDs, informing management decisions, and prognosis, [26][27][28] potentially further reducing the need for invasive testing.
This study is unique in that we utilized a large multi-centre, well characterized population, and applied standardized lung function values, which should improve accuracy of the lung function estimates.Limitations include the low numbers of patients that underwent SLB or that fulfilled the criteria for the pre-post-SLB group given the relatively stringent FVC requirement for cohort inclusion.However, the comparison of the SLB and non-SLB groups serves as a reasonable surrogate as their lung function change should presumably be similar if SLB had no effect.An important limitation is selection bias of patients who proceed to SLB in that these patients are presumably healthier, and in our cohort were younger yet had marginally lower baseline lung function.This study was designed to characterize the impact of SLB in patients with IPF given its known natural history of progressive disease.Findings may not be generalizable to non-IPF ILDs and the potential impact of SLB on lung function in these conditions warrants further study.We did not find an increased risk death or transplant in the SLB group, nor was there a significant increase in incidence of AE-IPF in the 2 years following SLB.The reasons for this are unclear, but may represent that in clinical practice, patients are being appropriately selected for SLB and that accelerated lung function decline is not occurring thanks in part to expert guidance on relative contraindications to biopsy and advances in surgical technique. 29Prospective studies would be required to fully address this question.Our study period overlaps with publication of the 2018 IPF guidelines, which may have changed practice with fewer patients referred for SLB.CARE-PF is comprised of tertiary care and specialized ILD centres which may also utilize highly experienced surgical sites for SLB, potentially limiting the generalizability of our findings.Finally, we cannot exclude survival bias in that patients who had SLB before referral to study sites but did not survive would not be accounted for while those who did survive contribute data to these retrospective studies.
In summary, we did not find that SLB was associated with accelerated lung function decline in patients with IPF, which may reflect appropriate patient selection for SLB and provide reassurance in scenarios where it is thought to be indicated.We also did not find an increased risk of death in this cohort recognizing the limitations of this analysis.However, our study did not evaluate other potential perioperative complications including pain, bleeding, infection or hospitalization.To minimize risks to patients, future work should develop non-invasive approaches to diagnosis, such as serum, genomic or radiomic biomarkers or to inform disease biology, activity, and response to therapy.
Abbreviations: DLCO, diffusion capacity of the lung; FVC, forced vital capacity; IQR, interquartile range; L, litres; n, number; SLB, surgical lung biopsy.a Deceased and transplant data presented from the 2 years of follow-up for the SLB and non-SLB group, and from the 1 year of follow-up after SLB for the pre-post-SLB group.In the SLB group one death was from acute exacerbation of IPF, one from lung cancer and the other of unknown cause.b Received either nintedanib or pirfenidone for at least 3 months during period of inclusion.

F
I G U R E 3 Mean annual FVC% change in the 12 months leading up to and 12 months after SLB in the pre-post-SLB analysis group.FVC, forced vital capacity percent predicted; SLB, surgical lung biopsy.

F
I G U R E 4 FVC% trajectory for pre-post-SLB analysis (n = 15) and by treatment status.FVC%, forced vital capacity percent predicted; n, number.

F
I G U R E 5 Transplant-free survival comparing the SLB and non-SLB groups.SLB, surgical lung biopsy.