Mechanical ventilation and histological fibrosis in patients with acute respiratory distress syndrome undergoing open lung biopsy


 Background: Mechanical ventilation is the most important support treatment for acute respiratory distress syndrome (ARDS); however, it brings with it the risk of ventilator-induced lung injury, which can lead to pulmonary fibrosis and prolonged mechanical ventilation. Our objective in this study was to examine the correlation between ventilator settings and histological fibrosis in ARDS patients. We also examined the impact of histological fibrosis on clinical outcomes and mortality.
Methods: We performed a retrospective analysis of patients with ARDS who received open lung biopsy in a tertiary care referral center in Taiwan between March 2006 and December 2019. Mechanical ventilator settings were recorded at the time of ARDS diagnosis and on the day of biopsy.
Results: A total of 68 ARDS patients who had received open lung biopsy with diffuse alveolar damage (DAD; the hallmark pathology of ARDS) were analyzed and stratified into non-fibrosis (n = 56) and fibrosis groups (n = 12). Mechanical power and airway pressures at the time of ARDS diagnosis and airway pressures on the day of biopsy were significantly higher in the fibrosis group than in the non-fibrosis group, and lung compliance on the day of biopsy was significantly lower. The duration of ventilator usage and time spent in the intensive care unit and hospital stay were all significantly higher in the fibrosis group. Hospital mortality was higher in the fibrosis than in the non-fibrosis group (67% vs. 57%, p = 0.748). In the fibrosis group, patients with AE-IPF (n = 5) received higher ventilator load and faced higher hospital mortality than DAD with fibrotic phase (n = 7) (80% vs. 57%, p = 0.242). A multivariable logistic regression model demonstrated that mechanical power at ARDS diagnosis and ARDS duration before biopsy were independently associated with histological fibrosis at open lung biopsy (odds ratio 1.493 [95% CI 1.014–2.200], p = 0.042; odds ratio 1.160 [95% CI 1.052–1.278], p = 0.003, respectively).
Conclusions: Our findings indicate that prompt action aimed at staving off injurious mechanical stretching of lung parenchyma and subsequent progression to fibrosis may have a positive effect on clinical outcomes.
Keywords: Mechanical ventilation, Acute respiratory distress syndrome, Open lung biopsy, Histology, Diffuse alveolar damage, Pulmonary fibrosis, Idiopathic pulmonary fibrosis, Outcomes


Background
Acute respiratory distress syndrome (ARDS) is a heterogeneous syndrome with complex pathophysiologic mechanisms characterized by severe hypoxemia and high mortality [1]. The pathogenesis of ARDS includes an exudative phase, a proliferative phase, and a brotic phase. Alveolar type II cell hyperplasia occurs after the initial exudative phase, resulting in the formation of resident broblasts and extracellular matrix. As disease progresses to the brotic phase, extensive basement membrane damage and inadequate or delayed reepithelialization can lead to the development of interstitial and intraalveolar brosis, which often requires prolonged mechanical ventilator support [2].
The typical histological hallmark of ARDS is diffuse alveolar damage (DAD) manifesting as hyaline membrane formation, lung edema, in ammation, and hemorrhage [3]. Roughly only half of the patients diagnosed with ARDS based on current criteria present evidence of DAD upon open lung biopsy (OLB) or autopsy [4][5][6][7]. Note however that not all ARDS patients progress to the brotic phase, as any imbalance in pro brotic or anti brotic mediators could affect progression to the broproliferative phase [8]. In patients with ARDS, pulmonary brosis often leads to prolonged mechanical ventilation and poor clinical outcomes [9,10].
Idiopathic pulmonary brosis (IPF) is a chronic, progressive brotic interstitial pneumonia with unknown etiology and poor prognosis. It is characterized by deterioration of the lung parenchyma structure and respiratory function [11]. Acute exacerbation of idiopathic pulmonary brosis (AE-IPF) has been de ned as an acute, clinically signi cant respiratory deterioration of unidenti able cause, commonly leading to acute hypoxemia respiratory failure requiring mechanical ventilation [12]. The typical histological features of AE-IPF are DAD and/or organizing pneumonia superimposed on usual interstitial pneumonia (UIP) pattern (i.e., acute lung injury occurring in an IPF/UIP lung), such that it shares many of the pathological features of ARDS [13,14].
A lung-protective ventilation strategy remains the cornerstone of treatment for ARDS patients and has been positively correlated with improved survival [2]; however, it brings with it the risk of ventilatorinduced lung injury (VILI) and subsequent lung brosis [15]. Researchers have yet to establish optimal ventilator settings for patients with pulmonary brosis. Our objective in this study was to examine the association between serial changes in ventilator settings and histological brosis in ARDS patients with DAD based on OLB and compare clinical outcomes between brosis and non-brosis groups.

Study design and patients
This retrospective study was based on analysis of all ARDS patients who underwent OLB at Chang Gung Memorial Hospital (CGMH) in Taiwan between March 2006 and December 2019. CGMH is a tertiary care referral center with a 3700-bed general ward and 278-bed adult intensive care unit (ICU). The exclusion criteria included age of < 20 years and histological ndings not indicative of DAD.
At our institution, the decision to perform OLB was made by the treating intensivist in cases where the etiology of ARDS was unknown and the patient presented rapid pulmonary in ltration following a complete microbiologic examination including bronchoalveolar lavage. Surgical procedures were performed by a chest surgeon in an operating room or at bedside in the ICU. Informed consent was obtained from the family prior to OLB. Under general anesthesia and mechanical ventilator support with adequate oxygenation, video-assisted thoracoscopic surgery or a 5-cm thoracotomy was used to secure the origins using an endoscopic stapler-cutter. The biopsy site was a new or progressive lesion identi ed via high resolution computed tomography (HRCT) or chest X-ray. Every tissue specimen was examined by pathologists. The local Institutional Review Board for Human Research approved this study (CGMH IRB No. 202000760A3 and 202100595A3), and the need for informed consent was waived due to the retrospective nature of the study.
Ppeak is equivalent to plateau pressure in pressure-controlled ventilation, and Ppeak was used as a surrogate for plateau pressure to calculate MP if not speci ed [17]. Ventilator-free days was de ned as the number of days between day 1 and day 28 or day 90 in which the patient breathed without assistance for at least 48 consecutive hours. Patients who did not survive to 28 days or 90 days were assigned zero ventilator-free days. Patients were strati ed into the non-brosis group (DAD with exudative or proliferative phase) or the brosis group (DAD with brotic phase or chronic feature of honeycomb brosis).

Data collections
Demographic data, comorbidities, and the etiologies of ARDS were recorded from hospital charts. The dates of hospital and ICU admission, mechanical ventilator initiation and liberation, date of ARDS diagnosis and OLB, ICU and hospital discharge, and time of death were collected. Arterial blood gas and mechanical ventilator settings parameters were recorded at approximately 10 a.m. daily after ARDS onset.

Histological diagnosis
Based on analysis of hematoxylin-and eosin-stained lung tissue slices, a diagnosis of DAD (with or without brosis) was made independently by at least two pathologists who were blinded to the patients' clinical information. Any discrepancies were discussed by the pathologists with the aim of reaching a nal consensus as to histological diagnosis. The presence of DAD indicated hyaline membrane formation, intra-alveolar edema, alveolar type I cell necrosis, alveolar type II cell proliferation, the interstitial proliferation of myo broblasts and broblasts, or organizing interstitial brosis [3]. AE-IPF represents DAD and/or organizing pneumonia superimposed on UIP pattern [13,14]. We de ned brosis as the manifestation of collagenous brosis or chronic appearance of microcystic honeycombing, or both [19].

Statistical analysis
Continuous variables are presented as mean ± standard deviation or median (interquartile range), and categorical variables are reported as numbers (percentages). A student's t-test or the Mann-Whitney U test was used to compare continuous variables among groups. Categorical variables were tested using the chi-square test for equal proportions or the Fisher's exact test. Risk factors associated with histological brosis at OLB day were initially analyzed using univariate analysis, followed by a multivariable logistic regression model with stepwise selection. The results are presented using odds ratio and 95% confidence intervals (CI). All statistical analysis was performed using SPSS 26.0 statistical software, and statistical signi cance was considered when a two-sided p value is less than 0.05.

Study populations
This study identi ed 89 ARDS patients who underwent OLB within the study period and screened for inclusion and exclusion criteria. All ARDS patients were deeply sedated and paralyzed during the initial phase, including the day of OLB, and most cases received pressure-controlled ventilation until attempts at weaning from mechanical ventilator. After excluding 21 patients who did not ful ll the histological DAD, 68 patients with histological DAD were analyzed (Fig. 1). Based on histological ndings, 56 patients were assigned to the non-brosis group, and 12 patients were assigned to the brosis group. In the brosis group, 7 patients had DAD with brotic phase and 5 patients presenting the AE-IPF.

Histological ndings
Compared to normal lung tissue ( Fig. 2A), DAD in the initial stage was characterized by the formation of hyaline membrane (Fig. 2B) and in later stage by interstitial brosis (Fig. 2C). The term honeycomb brosis refers to the typical appearance of cysts in scarred lung tissue (i.e., UIP pattern) (Fig. 2D).
Organizing pneumonia was deemed indicative of acute exacerbation of UIP (Fig. 2E).

Baseline characteristics and clinical variables: non-brosis and brosis groups
As shown in Table 1, we observed no signi cant difference between non-brosis and brosis groups in terms of age, gender, body weight, body mass index, or comorbidities. In both groups, most of the ARDS cases were attributable to pulmonary causes (n = 58, 85%). The median interval between ARDS diagnosis and biopsy was signi cantly longer in the brosis group than the non-brosis group (18  vs. 8 [5][6][7][8][9][10][11][12] days, p = 0.024). In terms of ventilator settings at the time of ARDS diagnosis, MP, Ppeak, and dynamic ∆P levels were signi cantly higher in the brosis group than in the non-brosis group (all p < 0.05). On the day of biopsy, Ppeak and dynamic ∆P values remained signi cantly higher in the brosis group than in the non-brosis group, and dynamic compliance was signi cantly lower (all p < 0.05).
The overall hospital mortality rate was 59%, and the mortality was relatively higher in the brosis group than in the non-brosis group (67% vs. 57%, p = 0.748). The duration of mechanical ventilation, length of ICU stay, and length of hospital stay were signi cantly higher in the brosis group than in the non-brosis group (all p < 0.05).
Baseline characteristics and clinical variables: DAD with brotic phase and AE-IPF groups As shown in Table 2, the patients in the AE-IPF group was older than those in the DAD with brotic phase group. There were no signi cant differences between the two groups in terms of age, gender, body weight, body mass index, or comorbidities. In terms of ventilator settings at ARDS diagnosis, we observed no signi cant differences between the two groups. The median interval between ARDS diagnosis and biopsy was longer in the DAD with brotic phase group than the AE-IPF group (28  vs. 7 [4][5][6][7][8][9][10][11][12][13][14][15][16][17][18][19][20][21] days). On the day of biopsy, patients in the AE-IPF group received higher MP, higher V T , higher Ppeak, higher mean airway pressure, higher dynamic ∆P, and higher RR than did patients in the DAD with brotic phase group, although the difference did not reach the level of signi cance (Table 2 and Fig. 3). Dynamic compliance was similar between the two groups on the day of biopsy.
Clinical outcomes: DAD with brotic phase and AE-IPF As shown in Table 3 and Fig. 4, 90-d hospital mortality was higher in the AE-IPF group than in the DAD with brotic phase group (80% vs. 57%, p = 0.242). The duration of mechanical ventilation, length of ICU stay, and length of hospital stay were higher in the DAD with brotic phase group than the AE-IPF group.
The number of ventilator-free days at day 90, and ICU-free days at day 60 were also higher in the DAD with brotic phase group than the AE-IPF group.
Factors associated with histological brosis at OLB After adjusting for signi cant confounding variables, a multivariable logistic regression model revealed that patients who had higher MP at ARDS diagnosis and had longer ARDS duration before biopsy were signi cantly associated with histological brosis at OLB (odds ratio 1.493 [95% CI 1.014-2.200], p = 0.042 and odds ratio 1.160 [95% CI 1.052-1.278], p = 0.003, respectively) ( Table 4).

Discussion
The primary insight in this research was the fact that ARDS patients with histological DAD and brosis based on OLB received signi cantly higher airway pressures, and had signi cantly longer duration of mechanical ventilator, longer length of ICU and hospital stay. Hospital mortality was higher in the brosis group than in the non-brosis group. In the brosis group, patients with AE-IPF received higher ventilator load and had higher hospital mortality than those in the DAD with brotic phase group.
DAD is the pathological hallmark of ARDS. In the current study, we enrolled only ARDS patients with histological evidence of DAD using data from OLB, unlike previous studies that included ARDS patients with and without DAD using data from OLB or autopsy [6, 20,21]. The pathogenesis of ARDS proceeds through an exudative phase (for roughly the rst week after ARDS onset), proliferative phase (between the rst-and third-weeks following ARDS onset), and brotic phase (beyond 3 or 4 weeks after ARDS onset) [2,22]. In the current study, the median duration from ARDS diagnosis to OLB was 28 days in the DAD with brotic phase group. Some of the patients had progressed to the DAD with brotic phase within 3 weeks of ARDS onset. This indicates that the diagnosis of ARDS depends on clinical criteria, and that histological brosis may begin before all the criteria for clinical diagnosis of ARDS are met [19]. In a multivariable logistic regression model, longer ARDS duration before biopsy were independently associated with histological brosis at OLB.
The causes of pulmonary brosis during ARDS progression are multifactorial (e.g., in ammation and VILI) [9,15]. The course of lung brosis can be traced to persistent injury and repair in response to mechanical strain and stress on epithelial cell resulting from volutrauma and atelectrauma [23]. MP refers to the energy delivered by a ventilator to the respiratory system per unit of time, as determined by volume, pressure, ow, and RR. Researchers have established that MP is of higher predictive value than individual ventilator parameters in assessing the risk of VILI [17,18]. Excessive MP has been shown to promote VILI and appears to be strongly correlated with histology (DAD score) and the expression of interleukin-6, a marker of in ammation [24]. Driving pressure is inversely proportional to lung compliance and aerated remaining functional lung size and has been linked to mortality in ARDS patients [16,25].
There were few studies to examine the correlation between serial changes in ventilator settings and the development of histological brosis in ARDS patients. Serial changes of ventilator settings may re ect the severity of nonresolving lung damage. In the current study, the MP and airway pressures (Ppeak and dynamic ∆P) received by the brosis group were signi cantly higher than those received by the nonbrosis group at ARDS diagnosis. Thus, it is likely that the formation of lung brosis can be attributed at least in part to energy load (i.e., MP). In a multivariable logistic regression model, higher MP at ARDS diagnosis was independently associated with histological brosis at OLB. Compliance of the respiratory system was associated with the severity of lung injury, duration of ARDS, extent of lung brosis, and clinical outcomes [19]. Dynamic compliance in the brosis group was signi cantly lower and dropped more rapidly than in the non-brosis group at biopsy day. This may be due to the presence of brosis and a longer interval between ARDS diagnosis and OLB in the brosis group.
IPF is a form of chronic brosing interstitial pneumonia characterized by progressive decline in lung function with radiological and/or histopathological indications of UIP [11,26]. Note however that UIP is not synonymous with IPF. The UIP pattern of brosis has also been linked to other conditions, such as connective tissue disease (mostly rheumatoid arthritis), drug toxicity, chronic hypersensitivity pneumonitis, asbestosis, and Hermansky-Pudlak syndrome [26,27]. AE-IPF and ARDS are quite similar in terms of DAD, lung in ammation, and respiratory mechanics. AE-IPF can lead to severe acute hypoxemic respiratory failure requiring mechanical ventilator support. Patients with AE-IPF face a higher risk of mortality (may reach to 95%) [14], which may be related to the fact that IPF patients tend to be older. In our study, ve of the twelve patients in the brosis group presented histological ndings indicative of UIP, and none of those patients presented with connective tissue disease, drug toxicity, or asbestosis. A pathologist and radiologist agreed that the cause of respiratory failure in that group was primarily AE-IPF. Patients in the AE-IPF group were older than those in the DAD with brotic phase group (mean age 73.6 vs. 59.3 years), and the mortality was higher than DAD with brotic phase group (80% vs 57%). Impaired lung mechanics due to structural, biochemical, and anatomical aberrations render brotic lungs susceptibility to VILI [28]. There is at present no solid evidence indicating the optimal ventilator settings for brotic lungs, including AE-IPF; some concepts can derive from the evidence regarding ARDS because both share some common features. A "lung resting strategy" to avoid high PEEP during expiration and thereby prevent further lung injury has been posited as an alternative to the "open lung approach" (for ARDS cases presenting only DAD) for patients with pulmonary brosis and UIP, due to the fact that the presence of brotic tissue renders the lung structure highly fragile and prone to VILI (i.e., "squishy ball lung" concept) [28]. In our study, patients in the DAD with brotic phase and AE-IPF groups received similar ventilator settings except for V T at ARDS diagnosis. However, patients in the AE-IPF group received higher energy load (i.e., MP), higher V T , and higher airway pressures than in the DAD with brotic phase group at biopsy day. It indicated that intensivists may not recognize the disease status well and apply lung-protective ventilation at ARDS onset promptly; however, as disease progression, patients in the AE-IPF group received higher ventilator load than those in the DAD with brotic phase group at biopsy day due to underlying chronic brotic lungs, which contributed to higher risk of VILI.
The strength of our study was that we investigated pulmonary brosis based on histological brosis from OLB. Previous studies examining the effect of pulmonary brosis on clinical outcomes in ARDS patients reported a link between HRCT scores indicative of broproliferative changes and clinical outcomes/mortality [10,29]. Nonetheless, thin-section CT scanning, including inspiratory, expiratory, and prone sequences, is the most important tool by which to evaluate pulmonary brosis. Serial images and quantitative estimates are also essential to differentiating brosis progression [27]. Overall, imaging alone cannot be relied upon to con rm destruction of the lung parenchyma, delineate active broproliferation, or degree of lung brosis [8]. At present, there is no de nitive biomarker for DAD, and HRCT ndings are insu cient to differentiate DAD from DAD with organizing pneumonia, which was indicative of acute exacerbation of UIP [30]. The only way to con rm the presence of DAD is to obtain lung tissues via OLB or autopsy [20]. Unfortunately, cases that end in autopsy are very likely more severe than live cases, and autopsy series are unable to differentiate clinical outcomes or effects on mortality [19,20]. In the current study, we investigated the effect of pulmonary brosis on clinical outcomes by enrolling ARDS patients with histological DAD and brosis who had undergone OLB.
This retrospective study was hindered by a number of limitations. First, all patients were from a single tertiary care referral center over a long enrollment period. Furthermore, we focused only cases of ARDS that had undergone OLB who ful lled the histological DAD with brosis (i.e., brotic phase or AE-IPF) or not, which limited the number of recruited patients. Note that we opted not to exclude the ve patients with chronic brosis (i.e., UIP pattern) from the brosis group (n = 12), similar to that of a previous ARDS study based on autopsies in which half of the brosis group (15 of 30 patients) also presented with chronic microcystic honeycombing [19]. Besides, we further divided the brosis group into the DAD with brotic phase or the AE-IPF group and compared clinical outcomes. Second, the causes of pulmonary brosis are complex and multifactorial, and the exact causal relationship between mechanical ventilation and pulmonary brosis was di cult to determine due to the retrospective nature of the study. Third, compliance with lung-protective ventilation tends to drop in real-world clinical practice over the long study period from 2006 to 2019. The ARDS patients included in this study received relatively high V T , which may have in uenced clinical outcomes. Finally, corticosteroids have anti-in ammatory and anti brosis effects; however, we opted not to address the use of steroid therapy, due to a lack of evidence pertaining to the bene ts of steroid treatment for persistent ARDS and brotic lungs.

Conclusions
Our ndings revealed that ARDS patients with histological DAD and brosis received signi cantly higher airway pressures, underwent mechanical ventilation for a longer duration and remained in the ICU and hospital for a longer period. Hospital mortality was higher in the brosis group than in the non-brosis group. In the brosis group, patients with AE-IPF received higher ventilator load and faced higher mortality than those in DAD with brotic phase.
Implementing optimal ventilator settings as early as possible may be necessary to reduce the risk of VILI and pulmonary brosis. Further large-scale studies will be required to identify the mechanisms by which mechanical ventilation induces pulmonary brosis, and de ne safety standards aimed at minimizing the risk of VILI and brosis. The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Ethics approval and consent to participate
The study was conducted in accordance with the guidelines of the Declaration of Helsinki and approved by the Institutional Review Board of CGMH IRB No. 202000760A3 and 202100595A3.

Consent for publication
Not applicable. Tables   Table 1 Background   Data was presented as mean ± standard deviation, count or median (interquartile range) AE-IPF acute exacerbation of idiopathic pulmonary fibrosis, ARDS acute respiratory distress syndrome, DAD diffuse alveolar damage, FiO 2 fraction of inspired oxygen, PaO 2 partial pressure of oxygen in arterial blood, PBW predicted body weight, PEEP positive end-expiratory pressure    Serial changes in ventilatory variables between ARDS diagnosis and open lung biopsy of patients in the DAD with brotic phase and AE-IPF groups. ARDS acute respiratory distress syndrome, DAD diffuse alveolar damage, AE-IPF acute exacerbation of idiopathic pulmonary brosis, MP mechanical power, V T tidal volume, PBW predicted body weight, PEEP positive end-expiratory pressure, Ppeak peak inspiratory pressure, ∆P driving pressure, RR respiratory rate, bpm beats per minute