Addressing Knowledge Gaps in the Early Detection of Bronchiolitis Obliterans Syndrome after Hematopoietic Cell Transplantation: An Official American Thoracic Society Research Statement

Abstract

Background

Bronchiolitis obliterans syndrome (BOS) is a late-onset noninfectious pulmonary complication of allogeneic hematopoietic cell transplant (HCT) that is often diagnosed at an advanced stage with severe lung impairment. Increasing use of HCT for the treatment of hematologic diseases worldwide translates to an increasing burden of BOS, particularly for the community pulmonologist. Early recognition of BOS, which offers the best opportunity to mitigate morbidity and mortality, is hampered by incomplete knowledge of the clinical course and disease process. The goal of this research statement is to survey our current understanding of BOS and to define the research agenda for the early detection of BOS.

Methods

We convened a multidisciplinary panel that included community representatives for an in-depth survey of the published literature followed by an online workshop.

Results

Major knowledge gaps were identified within interrelated themes of natural history and pathogenesis, risk factors, and the clinical diagnostic approach.

Conclusions

This statement reflects the detailed assessment of identified knowledge gaps with associated key research questions, as well as a proposed research road map to stimulate cross-disciplinary collaborations from preclinical to clinical investigations.

Contents

•  Introduction

•  Methods

•  Current Knowledge

  •  The Clinical Entity
  •  Pathogenesis of BOS
  •  Identifying Patients at Risk Clinical Approach
• Research Road Map

  • Considerations for Risk Factor Analyses
  • Pathogenesis and Early Detection
• Additional Gaps and Future Directions

  • Pulmonary–HCT Collaboration
  • A Distinct Clinical Entity
  • Interventional Trials

• Conclusions

Introduction

Bronchiolitis obliterans syndrome (BOS) and other late-onset noninfectious pulmonary complications (LONIPCs) affect as many as 20% of allogeneic hematopoietic cell transplant (HCT) recipients, contributing to significant morbidity and mortality for individuals who have been cured of hematologic disease (1, 2). BOS, the most common LONIPC, is characterized by new onset airflow obstruction due to the pathologic lesion of obliterative bronchiolitis (OB) and is a manifestation of chronic graft-versus-host disease (cGVHD), a systemic condition of alloimmune-mediated inflammation and fibrosis that results in end-organ dysfunction (3). The clinical impact of BOS after HCT includes chronic respiratory impairment, increased susceptibility to lung infections, and a 50–60% 5-year survival rate (4). Worldwide, approximately 39,000 allogeneic HCTs are performed yearly, and the incidence of HCT is increasing by 7% annually (5). Increasing numbers of HCT survivors who return to community care translates to an increasing burden of BOS.

Because small airways represent the “silent zone” of the lung, many individuals will have already developed advanced disease at the time of symptom onset and diagnosis. The widely used NIH definition of BOS is neither sensitive nor specific, which contributes to delayed clinical recognition (4, 6). Current treatment approaches, including corticosteroids, are largely ineffective in restoring lung function (7). In recent clinical trials of targeted therapies for refractory cGVHD (including ruxolitinib, belumosudil, and others), the response rate and quality of life were markedly lower for BOS than for other organ manifestations, reinforcing the notion that advanced lung dysfunction reflects irreversible structural changes of the airways (8–12).

Early recognition of BOS may offer the best opportunity to mitigate chronic lung morbidity and mortality (13, 14). A randomized controlled trial of the treatment of new-onset BOS with inhaled budesonide/formoterol and recent single-arm trials of systemic agents suggest that milder disease, possibly reflecting earlier recognition, can be halted or reversed (7, 15). Because of the rarity and insidious nature of the condition, our knowledge of BOS remains significantly incomplete. Gaps remain in fundamental knowledge of disease evolution (e.g., natural history), particularly in the contemporary era of evolving HCT practices that include newer antiinfective and GVHD prophylaxis and treatments. Without this knowledge, we cannot implement currently available therapies in a timely fashion, design novel therapeutic approaches, or fully evaluate the impact of early detection.

Insights can be drawn from the understanding of OB of other origins, particularly as a manifestation of chronic lung allograft dysfunction (i.e., CLAD-BOS), the syndromic term for chronic rejection following lung transplant, which shares a histologic similarity and a context of alloimmunity with HCT-BOS (16–18). Yet the syndrome manifests differently for HCT recipients, so the clinical approach should reflect the evidence base for the specific medical and biologic context of HCT.

This project was inspired by patients who experience the consequences of delayed diagnosis and our incomplete knowledge. The goal of this research statement is to survey our current understanding of early BOS and to define the research agenda for the early detection of BOS. This document is not intended to provide clinical practice and treatment recommendations. We aimed to identify knowledge gaps, emphasize key scientific research questions, and propose a research road map to address these gaps.

Methods

The co-chairs (G.-S.C. and A.B.) convened a multidisciplinary committee with individuals who met one or more of the following criteria: 1) pulmonologist or transplant hematologist with clinical expertise in LONIPCs; 2) investigator who has published original research on LONIPCs or airways diseases (specific expertise in immunology, airway biology, lung fibrosis, and/or GVHD was sought); 3) community pulmonologist who has encountered challenges of LONIPCs in their practice; and 4) patient or patient representative affected by BOS. See online supplement for additional details.

The chairs developed a conceptual framework around the following related themes: natural history and pathogenesis, risk factors, and clinical approach. The chairs designated a writing committee (A.S., J.T., K.M.W., and J.L.H.), who also served as group leaders. Committee members were assigned a group by topic based on expertise and area of interest; members were encouraged to participate in discussion on any topic. An initial literature search through PubMed was conducted with the aid of a medical librarian and then further refined by the group leaders. The literature survey was distributed to panel members for an initial assessment of current knowledge. A two-half-day online workshop was convened for formal talks to summarize the state of the science. Breakout sessions were conducted to identify key questions and recommended research road map. The final workshop agenda (see online supplement) as well as the identified gaps, key questions, and research road map represented in this document, were derived by consensus through online discussions and subsequent email communications.

Current Knowledge

The initial literature survey revealed the following general observations: 1) clinical epidemiology and natural history are derived mostly from retrospective single-center observational studies; 2) the vast majority of original investigations target BOS in the setting of lung transplant rather than HCT; 3) biologic insights are derived from preclinical studies in the setting of cGVHD and lung transplant; 4) translational studies are rare because of the paucity of human tissue; 5) very few preclinical or translational studies have a specific focus on incipient disease, so insights from early disease are derived from studies of established disease; and 6) no specific diagnostic biomarker has been clinically validated for early detection of BOS.

Here we summarize the current state of knowledge, gap areas, and key questions (Table 1).

Table 1.

Summary of Knowledge Gaps and Key Research Questions for the Early Detection of BOS after HCT

Themes	Gap Areas	Key Questions
The Clinical Entity	1.Natural history 2.Diagnostic definition	a.How should clinical phenotypes be defined, and how do these phenotypes correlate with pathogenesis and/or risk factors? b.How should the current NIH diagnostic criteria for BOS in patients with HCT be modified to enable earlier detection, while maintaining diagnostic accuracy?
Pathogenesis	3.Chronic GVHD pathways and alloimmune effectors 4.Early transplant events and acute lung injury 5.Lung microenvironment	a.What are the triggering events of BOS? b.Is there a mechanistic link between early posttransplant lung injury and the development of BOS? c.How does the bronchial epithelium orchestrate the immune response to lung injury and repair in the context of alloimmunity? d.At what point does alloimmune inflammation transition to irreversible fibrosis? e.Is large airway inflammation a manifestation of early disease pathogenesis? f.What is the role of endothelial injury in the development of BOS? g.What is the role of the microbiome in BOS pathogenesis? h.Do lung infections, specifically viruses, contribute to BOS pathogenesis as antigenic stimuli triggering alloimmunity, through direct epithelial injury, or both?
Identifying Patients at Risk	6.Risk factors 7.Role of viruses and lung infections	a.What risk factors for BOS are relevant in the current era of transplant practice? b.What combinations of cross-sectional and longitudinal risk factors can identify patients with high positive predictive value for BOS? c.Which specific viral and lung infections confer risk for the development of BOS? d.What is the “risk window” for BOS after a lung infection?
Clinical Approach	8.Screening 9.Diagnostic modalities 10.Biomarkers	a.How do we improve implementation of screening recommendations and early diagnostic testing, taking into account patient-centered factors? b.What clinical/physiologic/proteomic biomarkers identify patients with early BOS? c.Which tissues/organs should be targeted for early diagnostic biomarker discovery and at what longitudinal time points? d.What cells, cytokines, and other biomarkers in the BAL fluid identify early disease? e.Whom should we target for screening of BOS, and when should this screening begin? f.What is the optimal interval for lung function monitoring? g.Which novel PFT interpretative strategies, techniques of physiologic lung function assessment, and sensitive imaging techniques would allow for earlier detection of BOS? h.How should PFTs be used in association with clinical presentation and other modalities to diagnose BOS?

Definition of abbreviations: BAL = bronchoalveolar lavage; BOS = bronchiolitis obliterans syndrome; GVHD = graft-versus-host disease; HCT = hematopoietic cell transplant; PFT = pulmonary function test.

The Clinical Entity

Natural history and clinical phenotypes of BOS

The original descriptions of OB in bone marrow transplant recipients with severe, fixed airflow obstruction involved lung biopsy specimens demonstrating completely occluded terminal small airways, reflecting end-stage disease (14, 19). Classic obstructive physiology with severely reduced FEV₁ and reduced FEV₁/FVC ratio is associated with radiographic findings of air trapping and hyperlucency, a sign of hyperinflation (20, 21). Bronchial wall thickening or attenuation, mild bronchiectasis, and centrilobular nodules are common, but not universally seen, on computed tomography (CT).

Although BOS is anatomically defined by a pathologic lesion of the small airways, evidence suggests concomitant involvement of the proximal bronchial tree. Histologic examination of BOS explants from HCT and lung transplants demonstrate that airway obstruction can be seen starting from the fifth generation of the bronchial tree (22). The original descriptions of the clinical syndrome noted inflamed bronchi on bronchoscopy, and a bronchodilator response was seen in a subset of patients, indicating the involvement of airway smooth muscle and justifying the use of inhaled β-agonists for initial treatment (7, 14). Mild bronchial wall thickening as well as centrilobular micronodules and tree-in-bud abnormalities seen on chest CT at Day 100 after transplant were found to be a risk factor in a prospective cohort study, suggesting pan-airway involvement early in the disease process (1).

The incidence of BOS peaks within the first 2 years after HCT during the period of greatest risk for cGVHD, although presentations beyond 3 years do occur, often in the context of extrapulmonary cGVHD flares or recurrence (1, 23, 24). Disease course varies but is generally characterized by a relatively rapid decline in FEV₁ in the 6 months before clinical recognition, followed by relative stabilization (4). Whether stabilization represents a BOS-specific treatment effect versus natural history is often unclear given that the pretreatment pulmonary function test (PFT) trajectory is generally unknown, and lung function may have already reached a plateau at the time of clinical recognition. Rarely, patients present with a precipitous FEV₁ decline within 3–6 months of HCT, and severe disease is associated with earlier onset at <1 year posttransplant and a poorer prognosis (25–27). There are subgroups of patients who appear to experience a plateau with mild lung impairment, often in the context of treatment for cGVHD, whereas others continue to show progression to severe irreversible impairment despite treatment (4, 27, 28). BOS phenotypes, however, are incompletely characterized, and a method of early identification to distinguish rapidly progressive disease from a more indolent phenotype, and their correlative pathologies, has not been established. The notion that clinical outcomes are driven by different disease phenotypes is further supported by biopsy series: lymphocytic bronchiolitis presents identically to OB with new-onset airflow obstruction but has a much better prognosis (29). Whether lymphocytic bronchiolitis represents an earlier phase of the disease or simply a different phenotype of the syndrome remains unclear. The natural history of BOS suggests that optimal outcomes are most likely to be achieved if the disease is detected early in its course.

Current diagnostic definition

The NIH consensus diagnostic criteria for BOS, first proposed in 2005 and revised in 2014, provide a harmonized definition based on evidence of new airflow obstruction and the absence of infection (Table 2) (6, 30). These criteria represent the syndromic correlate of the pathognomonic OB lesion and have been widely adopted for clinical diagnosis, but have significant limitations. In a series of patients who underwent lung biopsy for suspected HCT-BOS, only 52% of the pathologically proven OB cases met the NIH spirometric definition; 24% of those that did meet NIH criteria were non-OB parenchymal entities (31).

Table 2.

2014 National Institutes of Health Chronic GVHD Consensus Diagnostic Criteria for BOS after Hematopoietic Cell Transplant

In the presence of a distinctive manifestation of chronic GVHD, the clinical diagnosis of BOS is sufficient to establish the diagnosis of chronic GVHD for the purposes of enrollment in clinical trials when all of the following criteria are met:
1.FEV1/vital capacity ratio <0.7 or the fifth percentile of predicted	A. Vital capacity includes FVC or slow vital capacity, whichever is greater
	B. The fifth percentile of predicted is the lower limit of 90% CI
	C. For pediatric or elderly patients, use the lower limits of normal, defined according to NHANES III calculations
2.FEV1 <75% of predicted with ⩾10% decline over <2 yr; FEV1 should not correct to >75% of predicted with albuterol, and the absolute decrease for corrected values should still remain at ⩾10% over 2 yr	—
3.Absence of infection in the respiratory tract documented with investigations directed by clinical symptoms such as chest radiographs, CT, or microbiologic cultures (sinus aspiration, upper respiratory tract viral screen, sputum culture, BAL)	—
4.One of the two supporting features of BOS	A. Evidence of air trapping by expiratory CT or small airway thickening or bronchiectasis by high-resolution chest CT
	B. Evidence of air trapping by PFTs: residual volume >120% of predicted or residual volume/TLC ratio increased beyond the 90% CI

Definition of abbreviations: BOS = bronchiolitis obliterans syndrome; CI = confidence interval; CT = computed tomography; GVHD = graft-versus-host disease; NHANES = National Health and Nutrition Examination Survey; PFT = pulmonary function test.

Adapted with permission from Jagasia and coworkers (6).

Importantly, the current spirometric criterion of a percent predicted FEV₁ (ppFEV₁) threshold of <75% is insensitive to early disease, particularly if a patient has a baseline ppFEV₁ well above the lower limit of normal (4, 32). Given that BOS is considered a small airway disease affecting the silent zone of the lung, incipient disease may not be reflected in FEV₁ until there is significant disease burden. Similarly, the requirement for a postbronchodilator ppFEV₁ to remain <75% is misleading because a bronchodilator response that improves, but does not fully restore, FEV₁ can be seen in some individuals, and even a nonsignificant improvement could fail this requirement (e.g., improvement from 73% to 76%) (7, 33). The criterion for airflow obstruction (FEV₁/FVC ratio <0.7 or lower limit of normal) also do not allow for recognition of a preserved ratio impaired spirometry presentation, as has been described in chronic obstructive pulmonary disease (COPD), and has been associated with small airway dysfunction (34–36). In HCT-BOS, this pattern may precede the fulfillment of NIH criteria (1, 27). Moreover, in those with preexisting lung diseases, the NIH criteria do not offer guidance on what degree of decline in lung function should raise suspicion for BOS as opposed to being ascribed to pre-HCT lung diseases, and, in these cases, the diagnosis of BOS requires a high index of clinical suspicion. Therefore, the NIH criteria may impede early recognition, which requires longitudinal frequent interval measurements and more sensitive metrics of lung function change.

In addition to suboptimal sensitivity for early disease, the 2014 NIH spirometric criteria lack specificity (31). Fulfillment of these criteria can be complicated by non-BOS conditions, including parenchymal lung disease such as organizing pneumonia, alveolar damage, and nonspecific interstitial pneumonia (31, 37–39). Although these entities have yet to be formally incorporated in the NIH definition of lung GVHD, they can confound recognition of BOS (40). Early identification to distinguish these coexisting entities from each other remains an ongoing clinical challenge. Moreover, the requirement for an absence of infection does not consider that infection can potentially coexist with and trigger BOS and thus contributes to a diagnostic odyssey that delays early treatment.

Summary and key questions

Major gaps in our understanding of the clinical entity that inform priority research questions include comprehensive characterization of BOS phenotypes and clinical–radiologic–pathologic correlations of early disease with anatomic airway involvement. The NIH definition is limited in sensitivity and specificity for early disease, reflecting our incomplete understanding of the clinical entity and additional forms of lung GVHD (Table 1).

Pathogenesis of BOS

In humans, the small airways are anatomically characterized by their small diameter (<2 mm), the absence of cartilage, a thin wall, as well as a position at the interface between external air and the bloodstream, making them particularly vulnerable to injury and obstruction (41). Although few studies have focused on early disease phases, knowledge gained from our current understanding of established disease will form the basis for developing biomarkers and screening and diagnostic approaches for early detection. We examined the current understanding of pathogenesis to highlight specific knowledge gaps, particularly those related to early posttransplant events and the lung microenvironment.

cGVHD pathways and alloimmune effectors in the lung

The presentation of BOS after HCT within the first 2 years after transplant is closely linked to the presence of immune dysregulation due to cGVHD, which likely explains the difference in timing of BOS after HCT compared with LT, in which BOS incidence increases over time. As a pulmonary manifestation of cGVHD, BOS likely shares pathogenic mechanisms with cGVHD in other organs following the three-step model: 1) acute inflammation induced by toxic insults, 2) alloreactive lymphocyte dysregulation and chronic inflammation, and 3) abnormal tissue repair and subsequent fibrosis. cGVHD involves expansion of donor-derived effector CD4⁺, CD8⁺, and B-cell populations and lack of adequate donor- or host-derived regulatory mechanisms, resulting in recruitment of effector cells to peripheral tissues (42–44). Impaired B lymphopoiesis is associated with increased B-cell activating factor levels in patients, which may support the survival of autoreactive and alloreactive B cells and their biased differentiation through plasmablasts rather than memory B cells, contributing to inflammation and fibrosis (45–52). In addition, an increased level of CXCL13 and a reduced number of circulating T follicular helper cells in humans support the model of the role of T follicular helper cells promoting germinal center and B-cell responses (53). Altered T cell reconstitution has also been described in humans, leading to an increased proportion of conventional CD4⁺ T cells from peripheral expansion and reduced T regulatory cell production (54). Granzyme K–expressing CD8⁺ T cells with increased expression of CCR5 have also been found to contribute to cGVHD-BOS in humans and mice (55). Skin and oral mucosal infiltration by Tbet⁺ Th1 and Th17 effector cells has also been identified in humans (56–58). Macrophages are also major cellular players in cGVHD-affected skin (59).

However, most of these particular immune dysregulation phenomena have yet to be linked to BOS specifically. Currently, few data confirm that mechanisms underlying BOS development are similar to those described in preclinical models. In a mouse model of cGVHD, the local formation of germinal centers and the deposition of donor B-cell alloantibodies in the small airways as well as macrophage infiltration were required for the development of lymphocytic bronchiolitis with few OB lesions (60–62). Murine models of OB, however, are limited by the lack of true terminal bronchioles that replicate human anatomy with fidelity, as well as the low penetrance of the disease (41).

Early transplant events and acute lung injury

The first events leading to BOS after HCT remain a critical question. It is hypothesized that there is a mechanistic link between early lung injuries in the peri- and posttransplant periods and the subsequent development of BOS. As suggested by epidemiologic studies, subclinical toxicities from the conditioning regimens such as busulfan and total body irradiation, as well as preexisting lung disease, may play a role in early pathogenesis (23, 63–65). In humans, early posttransplant lung complications encompass a spectrum of acute noninfectious presentations classified under the term “idiopathic pneumonia syndrome” (IPS), and experimental mouse models of systemic acute GVHD suggest the lung to be a target of acute GVHD (66). Lung injury may also be due to acute or occult pulmonary infections (66–68). The question of whether some antecedent injurious event mediated by alloimmunity, infection, or a combination thereof specifically predisposes to the development of BOS remains unresolved.

Insights into early events could be drawn from lung transplantation, in which acute rejection is well documented to increase the risk for subsequent CLAD-BOS (69, 70). However, although the occurrence of acute GVHD is associated with the development of cGVHD, this link has not been clearly established for HCT-BOS. No study to date has shown an association between IPS specifically and post-HCT BOS. IPS has a 1-year mortality rate as high as 70% and therefore represents a major competing risk for BOS that can potentially obscure any potential association (66, 71, 72). Yet, the notion of early posttransplant lung injury as a broadly defined factor in early BOS is supported by epidemiologic associations of early posttransplant spirometry declines and a history of respiratory illness or lower tract infection with the development of BOS (1, 64, 73). Moreover, acute cellular rejection is routinely monitored after lung transplant by sequential transbronchial biopsies, even in the absence of clinical symptoms. This is not the case for IPS, which is a clinical entity and is not routinely monitored or uniformly histologically characterized. Further studies are needed to determine whether subclinical IPS may occur and if IPS has an association with BOS similar to acute rejection after lung transplant.

Early posttransplant lung injuries share common pathways, such as acute inflammation with release of proinflammatory cytokines (including TNF-α [tumor necrosis factor-α] and IL-1β), epithelial and endothelial injury, and pulmonary recruitment of donor-derived macrophages and donor neutrophils (44, 72, 74, 75). An early robust inflammatory phase (due to toxicity, alloreactivity, and/or infection) may not be a prerequisite for subsequent fibrosis; persistent epithelial damage and subsequent “cross-talk” between epithelial cells, inflammatory cells, and fibroblasts in the context of endothelial cell activation may be sufficient for the development of fibrotic lung disease (76–78). Such a mechanism could explain why some patients with BOS do not have clinically significant antecedent lung inflammation; in the setting of imbalanced immune regulation, an allogeneic response to subclinical injury of lung epithelial cells could initiate a dysregulated reparative response, resulting in fibrosis of the airways.

One area of uncertainty is the role of lung and gut dysbiosis in the development of BOS. Reduction in diversity of gut microbiota associated with the use of broad-spectrum antibiotic agents was associated with the development of acute lung infiltrates in HCT recipients, as well as increased mortality (79). Pre-HCT differences in lung microbiome signatures have recently been associated with posttransplant acute lung injury in children (80, 81). The depletion of commensal microbiome constituents was found to be associated with pathogen enrichment, inflammation, fibroproliferation, and poor survival (82). Although the pre-HCT lung microbiome diversity inversely correlates with pre-HCT lung function, the contribution of the lung microbiome to the development of BOS remains unknown (81). Future studies should closely examine the association of the gut and pulmonary microbiome with BOS risk because it is possible that the resident microbiome influences the degree of airway inflammation (83).

The lung microenvironment: role of airway epithelium and small airway fibrosis

Effective repair after an injurious insult requires a concert of cell types, including epithelial cells, fibroblasts, and macrophages (84). Because damage to the bronchiolar epithelium is hypothesized to be a trigger for BOS, it is relevant to focus on functional alterations in bronchial epithelial cells. In addition to being a mechanical barrier, normal airway epithelial cells are also key sensors for the environment, orchestrating immune and structural cell responses to airway injury (85). The main cell types lining these small airways include secretory club cells (CCs) and basal cells (86–88). The initial repair and maintenance of small airways is likely mediated by these nonciliated CCs and basal cells that serve as progenitors for ciliated and secretory airway populations, including themselves. CCs secrete distinctive proteins, including CC 10-kD secretory protein (CCSP) and other secretoglobins, mucins, antimicrobial proteins, and surfactants that contribute to homeostasis and host defense (89–92).

There is evidence that the composition and function of the airway epithelium are altered in OB: CC loss may play a key role in the pathogenesis of CLAD-BOS by increasing the pulmonary adaptive/humoral response (93–96). CCSP, as a marker of CC loss, has been shown to be lower in the serum of HCT recipients with BOS and in the BAL fluid of patients with CLAD, although it should be noted that CCSP levels can vary with inflammation (94, 97, 98). Reduced CCSP in a knockout murine bone marrow transplant BO model resulted in decreased lung compliance, higher degree of fibrosis, and increased inflammatory markers (99). In a transgenic murine model of inducible and sustained CC injury, the loss of CCs is characterized by the release of inflammatory mediators including TGF-β (transforming growth factor-β) and the accumulation of alternatively activated macrophages, which led to the development of constrictive obliterative airway lesions (95, 100, 101). TGF-β is a canonical profibrotic cytokine that signals proximal airway fibroblasts and allows their differentiation to α-SMA (α-smooth actin)–expressing myofibroblasts, promoting collagen and extracellular matrix protein deposition and airway remodeling (101, 102). In contrast, models of other fibrotic lung diseases show that CC deletion results in inhibition of the TGF-β response, blockade of epithelial–mesenchymal transition, and reduced subepithelial myofibroblast expansion, suggesting plasticity or different phenotypes of these cells (93, 103, 104).

Data from lung transplant recipients show that alteration of basal cells is central to the pathogenesis of CLAD-BOS in that these cells, as well as CCs, are the target of cytotoxic CD8 T lymphocytes. A parallel pathology has not yet been shown in HCT-BOS (95, 105, 106). In other diseases associated with airway fibrosis, it has also been shown that basal cells can exhibit a profibrotic profile, stimulating fibroblast proliferation and collagen production (106). In BOS, it is likely that injury to a regenerative population (CCs or basal cells) contribute to the formation of myofibroblasts and subsequent fibrosis.

The detection of lung epithelial cell chimerism in patients with OB after HCT or lung transplant suggests that marrow cells contribute to the regeneration of the bronchiolar epithelium (107). Experimental data from a mouse model demonstrate that nonhematopoietic cells are the main source of this chimerism (108). Better characterization of the cells involved and the study of their interaction with the lung microenvironment in the pathophysiology of early BOS merits further investigation. Finally, an emerging concept of inflammatory memory acquisition by epithelial stem cells may apply to lung biology, and its contribution to BOS pathogenesis should be considered (109).

Macrophage activation is likely to play a key role in the progression of fibrotic lesions in OB. In response to local cytokines and microbial signals, macrophages can endorse a wide spectrum of phenotypes with different functions (e.g., from proinflammatory M1 phenotype to antiinflammatory and prorepair M2 phenotype). In a mouse model of cGVHD, donor macrophages mediated bronchiolar lesions in an IL-17 and CSF-1/CSF-1 receptor–dependent manner by disrupting the balance of M1/M2 macrophage function (62, 110). In the same murine model, administration of pirfenidone, a pulmonary antifibrotic agent that blocks TGF-B, suggests that lung GVHD is mediated by donor M2 macrophage infiltration and excessive TGF-β production (110). In a single-arm clinical trial of HCT-BOS, pirfenidone was associated with improved cGVHD indices as well as FEV₁ improvement in a subset of patients (111, 112). The analysis of lung explants from patients undergoing lung transplant for post-HCT BOS found donor-derived M1-macrophages in early-stage lesions of BOS, whereas M2 polarization was found in the late-stage lesion (113). The role of human macrophages in the development of fibrotic lesions was confirmed in skin GVHD and more recently by the results of the phase II clinical trial evaluating an anti-CSF-1R antibody, axatilimab, in steroid-refractory cGVHD (10, 59). Macrophage-specific cathepsin B expression reflecting excessive tissue remodeling and persistent inflammation is observed at significantly higher levels in the BAL fluid of patients with CLAD compared with healthy donors (114).

Taken together, these key studies have identified a complex interplay among multiple cell types that orchestrates a series of pathologic events representing inflammatory and fibrotic pathways that may occur in tandem or independently of one another in some instances. The relevance of these pathways in early disease pathogenesis of HCT-BOS requires further study.

Summary and key questions

The pathogenesis of BOS is hypothesized to be due to multiple lung insults starting before, at the time of, or after transplant, with a key contribution of dysregulated immune milieu at the time of injury (e.g., involvement of alloreactive T cells) and subsequent remodeling of the bronchial epithelium (Figure 1). Although injury to key small-airway cell types, such as CCs, appears to be a pivotal event mediating inflammation and fibrosis, the timing of these processes in patients with BOS remains unclear. Major knowledge gaps include the sequence of antecedent lung injury events and when the early disease process becomes irreversible (Table 1).

Figure 1.

Proposed pathogenesis of hematopoietic cell transplant–associated bronchiolitis obliterans syndrome (BOS). Current knowledge of the pathogenesis of BOS comes from humans (top panel) and a mouse model of chronic graft-versus-host disease that develops lymphocytic bronchiolitis with few obliterated bronchioles (bottom panel); specific aspects of pathogenesis are reviewed in References 41–43, 65, and 71. See online supplement for additional references. BOS pathogenesis, which shares similarities with those of systemic chronic graft-versus-host disease, usually includes acute inflammation, alloreactive T and B lymphocyte dysregulation, chronic inflammation, and abnormal tissue repair, and subsequent airway fibrosis, although inflammatory pathways and fibrotic pathways may occur independently of one another. In predisposed patients, successive insults to the bronchiolar epithelium causing acute inflammation, such as conditioning or viral infections, are hypothesized to be a trigger for BOS. Alteration of epithelial progenitor cells, particularly club cells and basal cells, may be associated with the onset of BOS. T-cell dysregulation includes increases in cytotoxic CD8⁺ T cells and CD4⁺ TH17 cells and a decrease in T regulatory cells. T follicular helper cells are crucial for the activation of germinal-center B cells in the production and deposition of pathologic immunoglobulin and collagen, ultimately causing BOS. B-cell dysregulation is suggested by increased B-cell activation factor as well as specific subtypes of B cells. Macrophage recruitment and activation play a key role in the progression of fibrotic lesions in obliterative bronchiolitis. Donor macrophages mediated bronchiolar lesions in an IL-17 and CSF-1/CSF-1R (colony-stimulating factor 1 receptor)–dependent manner by disrupting the balance of M1/M2 macrophage function. Increased levels of TNF-α were found in the BAL fluid of patients at various stages of the process, whereas TGF-β signals proximal airway fibroblasts and allows their differentiation to myofibroblasts, promoting collagen and extracellular matrix protein deposition and airway remodeling. Ab = antibody; BAFF = B-cell activating factor; CCSP = club cell secretory protein; DAMP = damage-associated molecular pattern; DC = dendritic cell; ECM = extracellular matrix; GC = germinal center; MMP9 = matrix metalloproteinase 9; PAMP = pathogen-associated molecular pattern; PDGFα = platelet-derived growth factor receptor α; RV = respiratory virus; Tfh = T follicular helper cell; TNF-α = tumor necrosis factor-α; TGF-β = tumor necrosis factor-β; Treg = T regulatory cell.

Identifying Patients at Risk

Patients at high risk for HCT-BOS should be identified for more intensive monitoring or preemptive therapies. Current data are limited by the low incidence of BOS and the tendency toward retrospective single-center studies, such that modeling strategies may overstate causal inference assumptions. Here we describe known risk factors for HCT-BOS in relation to the transplant procedure, which can be divided temporally into pre-HCT (comorbidities present >1 mo before HCT), peri-HCT (within 1 mo before and after HCT, encompassing factors related to the transplant), and post-HCT (clinical events and measurements >1 mo post-HCT).

Risk factors

Pretransplant comorbidities

Preexisting airflow limitation or obstructive lung disease is a well-characterized risk factor that is commonly seen in HCT recipients with asthma or a tobacco smoking history (23, 65, 73). However, the mechanisms whereby pre-HCT airway diseases influence BOS risk are not clear; it is possible that preexisting airway epithelial metaplasia or inflammation may be more likely to trigger cGVHD in the lung if a subsequent injury occurs. It must be considered that, for some patients, a lower pretransplant FEV₁ and/or FEV₁/FVC ratio (even if in the normal range) can make it more likely that diagnostic criteria for BOS are reached, essentially creating a form of verification bias (“diagnostic criteria bias”) without the necessity of a causal relationship (115).

Older recipient age is a risk factor for BOS, but it is uncertain if this risk exceeds the risk of increased age for cGVHD in general (116–120). Possible age-related mechanisms include increased risk for organ injury with reduced ability to repair, the lack of thymus reconstitution to permit a diverse de novo T-cell repertoire, and a greater likelihood of use of older donors with oligoclonal T-cell populations rather than alternative donor sources such as cord blood that have overall lower rates of cGVHD (43).

Finally, several studies have identified that male recipients receiving female donor transplants may be more likely to develop HCT-BOS (1, 64, 117). The exact mechanisms that drive this increased risk are unclear; the increased reactivity of T and B lymphocytes from female donors due to estrogen receptor expression, which are lacking in male lymphocytes, has been postulated (121, 122). Moreover, H-Y antigens, being foreign to female donors, have been identified as potential targets (123).

Peritransplant

The primary risk factors related to the transplant procedure that influence the risk for subsequent BOS are the use of myeloablative conditioning regimens (especially those that are busulfan-based), peripheral blood stem cell source, and unrelated donors (63, 64, 124–126). The use of unrelated donor transplants vastly expands the pool of possible donors but is associated with an increased risk of alloimmunity (127–129). It is likely that the higher risk for BOS with unrelated donors and peripheral blood HCT parallels the overall increase in cGVHD rates with these donor cell sources (130). Conversely, the use of agents that reduce the rate of cGVHD in general, such as in vivo T-cell depletion with anti-thymocyte globulin or alemtuzumab, reduced the risk for BOS (63, 131, 132). Emerging data suggest that this may also be the case for posttransplant cyclophosphamide (133, 134).

Posttransplant

Extrapulmonary cGVHD is the most well-documented risk factor for BOS; approximately 5–10% of all allogeneic HCT recipients will develop BOS, but the incidence increases to nearly 15% among those who have evidence of extrapulmonary cGVHD (23, 116–118, 125). Furthermore, the cause-specific hazard of BOS was increased 2.9-fold (95% confidence interval, 0.98–8.60) after the occurrence of a chronic GVHD in a prospective cohort (1). Presumably, systemic alloimmunity is a requirement for pulmonary cGVHD, but most people with extrapulmonary cGVHD do not develop BOS. Early pulmonary complications after HCT, which include lung infections, appear to also confer increased risk (1, 63–65).

Multiple studies have demonstrated posttransplant decline in spirometric parameters as a risk factor for BOS. In a prospective European study, an absolute decline of ≥10% in FEV₁ at Day 100 conferred a significant risk of later BOS in a multivariable analysis. As previously noted, this study also showed that bronchial wall thickening on chest CT at Day 100 was associated with BOS and could reflect incipient disease (1). In a retrospective analysis of 180 patients with BOS, decreases in midflow rates (forced expiratory flow between 25% and 75% of FVC [FEF_25–75]) at posttransplant Day 80 compared with pretransplant baseline had a greater impact on the risk of developing subsequent NIH BOS than declines in FEV₁ (73). This early posttransplant spirometric risk supports the notion that these physiologic changes reflect early disease, and, relatedly, that overt or subclinical posttransplant lung injury events (e.g., aspiration and inhalational exposures) play a role in disease pathogenesis. However, the positive predictive value of new spirometric declines is low, and most declines not meeting BOS criteria are due to other conditions (135).

The role of viruses and lung infections

Risk factors linked to immune function have been identified for HCT-BOS and are consistent with the idea that lung infections are triggering events for the development of BOS. Immunoglobulin deficiency, suggesting an increased susceptibility to infection, has been reported as a risk factor, with risk increasing at an IgG threshold of <350 mg/dl (23, 25, 136). Impaired innate immunity was also associated with the development of BOS. Genetic polymorphisms in Toll-like receptors that recognize bacterial and viral pathogens as well as in bactericidal/permeability-increasing protein have been associated with the development of posttransplant BOS (137, 138). Recent human data revealed that specific minor antigen expression in the lung could predict the risk of developing HCT-BOS, suggesting that BOS is not only triggered by external inflammatory factors such as viral infection, but also by intrinsic predisposition associated with minor antigen expression (139).

Cytomegalovirus (CMV) has long been associated with an increased risk of cGVHD but rarely appears as a risk factor in older models focused on BOS (140, 141). In a contemporary cohort, first-onset herpesvirus reactivation (including CMV, human herpesvirus 6, and Epstein-Barr virus) was an independent risk factor for early acute lung injury syndromes after adjusting for age, HLA mismatch, and cord blood source for HCT; CMV reactivation increased the risk for subsequent BOS by 70% (142). This is consistent with data from a prospective lung transplantation cohort study in which CMV viremia was associated with a 61% increase in the risk of future CLAD development (143). It is possible that latent viral reactivation in the early posttransplant period promotes an inflammatory milieu that primes the lung for subsequent chronic lung injury (144, 145). Newer CMV prophylaxis agents may significantly mitigate the BOS risk for this specific virus, and therefore CMV as a risk factor should be reexamined in this context.

Community-acquired respiratory viral infection (RVI) is a well-established risk factor for CLAD-BOS in lung transplantation (146–152). In HCT recipients, RVIs have long been associated with poor outcomes in HCT recipients and are presumed to be a major risk factor for BOS (153–155). However, to date, only two single-center retrospective studies have been conducted in adults to demonstrate RVI as a risk factor for posttransplant airflow decline (156, 157). In a pediatric HCT cohort, RVI early after transplant was associated with early acute respiratory failure (i.e., IPS) as well as BOS (158). Although this association has most commonly been attributed to parainfluenza or respiratory syncytial virus lower tract infection, it is probable that most RVIs, including severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), contribute to BOS development, although the risk for progressive airway injury may vary (157–159). An area of uncertainty is the duration of the “risk window” for BOS after any RVI, given that RVIs can cause transient lung impairment, but most RVIs will not lead to BOS. In studies of CLAD, the magnitude of the association between RVI and CLAD decreases as the length of time between these events increases (146).

RVI likely represent an additional injury superimposed upon background airway inflammation, a phenomenon that is evident in asthma and other airway diseases (160, 161). In HCT-BOS, it is possible that RVIs serve as antigenic stimuli for alloimmune attack, as well as causing direct epithelial airway injury (158). In the setting of subclinical antecedent lung injury (e.g., early after transplant), RVI (and possibly bacterial infections) may further perturb an abnormal bronchial epithelial milieu and induce further inflammation and metaplasia in a way that is self-propagating, as has been described in COPD, but direct mechanistic studies are lacking (162). Resolution of inflammation and injury following RVI may be crucial to prevent the development of airway remodeling that eventually leads to BOS.

Lower respiratory tract infection in the first 100 days after transplant is a described risk factor for the development of BOS (1). To date, no specific bacterial or fungal organism has been linked as a trigger for BOS, but lower tract bacterial infection is common in patients with established BOS and associated with progression of FEV₁ decline (163). There is a high prevalence of occult infection in the BAL fluid of patients with BOS enrolled in clinical trials, but it is unclear whether these infections represent culprits or bystanders in BOS development (164).

Summary and key questions

Identifying HCT recipients at high risk for BOS is challenging because of the low incidence of BOS; the positive predictive value of known clinical risk factors is inconsistent. Respiratory infections, particularly RVIs, are likely a key risk factor, but these are also common nonspecific events after HCT. Key questions include how to develop a clinically useful risk assessment for early detection (Table 1).

Clinical Approach

The current NIH BOS definition and the lack of a clear definition for early disease challenge the clinical approach to the early detection of BOS. Unlike lung transplant recipients in whom the lung is the underlying chief medical concern for their ongoing medical care, allogeneic HCT recipients must contend with more common non-lung medical issues, including cGVHD of other organs and the risk of hematologic relapse. This clinical reality poses challenges for the implementation of screening strategies and early diagnostic tools.

Screening for BOS

Unlike skin or mucosal GVHD, the advent of BOS is clinically silent. It is currently assumed that the “onset” of disease is the time of diagnosis; however, with BOS, symptom-driven PFT diagnosis captures a disease state that is already well established. Despite nearly two decades of society-endorsed screening recommendations, uptake of routine PFTs at many HCT centers remains poor, with only 30–40% of allogeneic HCT recipients undergoing routine posttransplant PFTs (4, 165, 166). These data corroborate observations among the committee members that many HCT physicians at their centers are not routinely ordering PFTs after the early posttransplant period. Although optimal screening intervals still need to be determined based on a more complete understanding of BOS natural history, implementation of screening by the transplant caregiver remains a major clinical gap because of the perception that PFTs are a burden. The recent publication of clinical guidelines for the early detection of BOS in pediatric HCT recipients reflect the evidence base behind current clinical practice, but broad implementation of these guidelines remains to be seen (167).

Recently updated general posttransplant care recommendations include PFT screening every 3 months for the first year after HCT, every 6 months for the second year, and then annually for 5 years after HCT (168). The 2020 NIH cGVHD consensus project recommended that, for patients with cGVHD, the cadence of screening should be every 3 months with spirometry for ≥1 year after diagnosis and annually with a full PFT (which includes spirometry, lung volumes, and Dl_CO), along with spirometry or PFTs every 3–6 months until systemic immunosuppressive therapy has been weaned (32). Short-interval repeat testing, e.g., 2–4 weeks, is recommended for a decline in the FEV₁ of ≥10% from the patient’s baseline or Day 100 assessment (32). Recent pediatric guidelines also reinforce these screening recommendations (167). Additional high-risk categories, such as cases of recent RVI, have been suggested for more frequent interval testing, but high risk is loosely defined, and those with preexisting airflow obstruction, Day 100 FEV₁ impairment, or recent lower tract respiratory infection have not been formally incorporated into screening recommendations (17, 135, 169, 170).

At present, conventional PFTs remain the primary screening modality for HCT-BOS. Portable or handheld spirometers offer the possibility of frequent interval lung function monitoring at home without the burden of going to a PFT laboratory. Remote monitoring by a care team is now possible with Bluetooth-enabled spirometers that synchronize to a patient’s smartphone or tablet device and transmit measurements to a cloud-based portal. Although home spirometry is routine for monitoring of acute rejection in lung transplant recipients, allogeneic HCT recipients often have significant nonpulmonary medical concerns that may limit adherence. Two prospective studies to test the feasibility of wireless home spirometry in HCT recipients at thrice-weekly to weekly intervals demonstrated reasonable adherence (>70% of required measurements) (171, 172). In a highly selected population with cGVHD, weekly home spirometry was able to detect FEV₁ decline antecedent to the clinical diagnosis in some cases of BOS (172). Clinical implementation of remote spirometric monitoring will require attention to adherence, determination of an actionable FEV₁ change threshold, and an associated care and communications plan.

Diagnostic modalities for early detection

PFTs

As noted earlier, subtle spirometric changes not meeting the defined 2014 NIH ppFEV₁ threshold (<75% predicted, FEV₁/FVC ratio <0.7/lower limit of normal) could identify patients with BOS earlier in the course of the disease (6). Retrospective data have supported that a decrease in midflow rates was associated with subsequent obstructive lung disease; a change in FEV₁ (e.g., a decrease of 10%) may be predictive of BOS but has poor predictive performance (73, 166, 170, 171, 173). Of 1,325 patients with “pre-BOS” (defined as a 10% decrease in FEV₁ and/or a 25% decrease in FEF_25–75 vs. pre-HCT baseline) in a contemporary retrospective HCT cohort, only 72 went on to develop NIH BOS (5%) (135). Combining the presence of pre-BOS with clinical factors, including active cGVHD, modestly improved the positive predictive value (135).

PFT interpretive strategies specific for small airway disease in any medical context have not been defined (174). Detection of subtle pulmonary function changes is challenging against the backdrop of normal day-to-day biological variability, particularly for measures such as FEF_25–75 that exhibit especially high variability and poor reproducibility. Therefore, the emphasis for interpretation should be on the longitudinal trend of these parameters (174). In the NIH BOS definition, the longitudinal change metric for FEV₁ of 10% from baseline or in the 2 years before has been assumed to be a significant change indicating disease, but this has not been validated. Additional spirometric parameters that may be more sensitive measures of early obstructive physiology, such as FEV₃/FVC ratio and the application of z-scores, have yet to be applied to BOS definitions (174). PFT screening intervals must reflect the tempo of subclinical changes that presage BOS to ascertain trends. An appropriate spirometric marker to trigger early diagnostic workup for BOS remains to be determined.

Other PFT parameters, including those obtained by body plethysmography, that implicate increased airway resistance or a reduction in flow rates in the peripheral airways include an increase of residual volume/TLC ratio or functional residual capacity/TLC ratio, neither of which has been studied in BOS (174). Because hypoxemia is not a feature of early disease, and BOS does not directly affect the parenchyma, Dl_CO is less relevant for early detection. Posttransplant Dl_CO impairments are common, nonspecific, and not associated with BOS risk (135, 175).

Additional techniques that assess distal airway physiology include multiple breath washout (MBW) and oscillometry (impulse or forced oscillometry techniques). These technologies may overcome the effort-dependent variability of spirometry and be more sensitive in detecting changes in small airway resistance (176). However, their role in HCT-BOS remains investigational. MBW measures nitrogen clearance curves with a gas analyzer, providing a measure of ventilation distribution and gas clearance (lung clearance index [LCI]), partly dependent on small airways. LCI correlates with active BOS and spirometry metrics in pediatric HCT studies (177, 178). In an adult HCT population, nonincreased LCI was shown to have a high negative predictive value but with the significant limitation of poor specificity (179). Oscillometry, which measures airway impedance, has been shown to correlate with spirometry but failed to show predictive value for early BOS in a lung transplant population (180). Similarly, in an exploratory study in children who had undergone HCT, MBW failed to diagnose earlier BOS compared with spirometry (177). Based on these data, these techniques have no clear advantage over spirometry except in pediatric patients aged <8 years in whom spirometry cannot be reliably performed (32). However, more studies are required to fully evaluate their utility, especially considering their low cost and relatively low resource use compared with conventional laboratory-based PFTs. Spirometry is widely available and accepted as a screening test for BOS but is limited by lower positive predictive value; therefore, a newer technology would be useful only if it improved the positive predictive value and increased specificity in clinical assessments. To date, no new technology for the detection of pulmonary impairment has proven to be clearly superior to spirometry, but these techniques have the potential to be additive to spirometry in assessing small airway disease, as in asthma (181).

Chest Imaging. Quantitative CT

Chest CT scans are routinely used clinically in conjunction with PFTs to make a diagnosis of BOS. Quantitative imaging modalities corroborating the obstructive physiological impairment seen in small airway disease are of considerable interest to increase screening and diagnostic sensitivity. For example, density-mapping CT techniques in the lung transplantation population demonstrate areas of low attenuation that predate the diagnosis of CLAD-BOS on the order of years (182). Parametric response mapping (PRM) is an automated software algorithm developed initially for COPD and/or emphysema that uses a ventilation map (i.e., densitometry) on CT images to quantify functional small airway disease and gas trapping (183). PRM signatures have been validated in BOS; PRM-derived small airway disease was found to correlate with ppFEV₁ and FEV/FVC ratio, with measurements remaining consistent between site-specific CT acquisition protocols (184). Compared with qualitative radiologic features of BOS (mosaic attenuation, air trapping, centrilobular ground-glass nodules, and bronchial wall thickening or dilation), the combination of radiologist assessment and quantitative lung density analysis demonstrated improved diagnostic accuracy for HCT-BOS compared with either method alone (185). It remains to be determined whether routine implementation of these quantitative imaging techniques is feasible.

Machine learning on quantitative CT features.

For individuals after HCT, unsupervised machine learning (ML), in which the algorithm has no information about patient labels, has been applied to quantitative CT (qCT) of the chest to distinguish successfully between individuals with normal lungs, mild BOS, and severe BOS (186). Supervised ML classification models, in which the algorithm attempts to diagnosis individuals with BOS, have found thresholds for inspiratory and expiratory lung volumes combined with quantitative air trapping that identify HCT-BOS with relatively high specificity and sensitivity (186). The application of ML models to the analysis of qCT strain metrics (a reflection of dynamic lung parenchymal distortion during breathing) was able to distinguish early BOS from NIH-BOS and early BOS from normal lungs with a high degree of sensitivity and specificity (187). In CLAD-BOS, ML models have also identified early disease (188, 189). To date, investigations of BOS with ML have yet to be extended beyond radiomics to other diagnostic modalities such as PFT parameters or circulating biomarkers.

Quantitative magnetic resonance imaging.

Quantitative magnetic resonance imaging (MRI) is of particular interest in visualizing and quantifying lung structural–functional abnormalities at high temporal and spatial resolution to gain information regarding regional disease involvement. Furthermore, techniques such as diffusion-weighted ¹²⁹Xe MRI generate quantitative diffusion metrics that can inform lung microstructure beyond the airways (190). There are data supporting superior sensitivity and specificity for early lung structural–functional impairment from small airway disease in asymptomatic smokers and ex-smokers without spirometric impairment (191). MRI techniques in lung allografts demonstrated the ability to detect early CLAD-BOS using an oxygen transfer function (192). Data from a population of pediatric allogeneic HCT recipients found that ¹²⁹Xe MRI ventilation deficits were detected in patients with normal spirometry (i.e., ppFEV₁ >80%), supporting ¹²⁹Xe MRI as a sensitive imaging biomarker, but, to date, no study has proven that ¹²⁹Xe MRI can detect early BOS more accurately than spirometric methods (193). Prospective data assessing functional and regional changes in an adult population are being collected (clinicaltrials.gov identifier NCT04029636).

Bronchoscopy and lung biopsy

Given the high prevalence of occult infections in suspected BOS, fiberoptic bronchoscopy with BAL analysis to rule out infection is recommended as part of the diagnostic workup (167). Currently, the NIH BOS diagnosis is contingent on the absence of active lung infection (6), but fixed airflow obstruction can persist despite antiinfective therapy, suggesting that infections are concurrent with and likely promote disease progression (81, 164, 194–196). The impact of emerging metagenomic assays on the diagnosis of infection remains to be seen. Basic knowledge regarding the composition of BAL fluid in HCT recipients outside of the context of acute lung injury is extremely limited, and it cannot be assumed that the lung microenvironment is the same as that in a healthy individual who has not undergone HCT (197). In a small series of HCT recipients with new late-onset airflow obstruction evaluated with bronchoscopy, two patterns emerged in the absence of an identified infection: a neutrophilic- or lymphocytic-predominant cell count (198). These observations have not since been reproduced systemically in the era of improved infectious diagnostics and formal NIH consensus definitions, which were first proposed in 2005 (30). In a single-arm trial of montelukast for the treatment of established BOS, BAL analysis performed to study immune correlates showed a relatively bland cellular composition in the absence of infection, but further analysis showed that the CD4⁺ lymphocytes were predominantly Th2 and that all immune cell lineages expressed leukotriene receptors (195). Aside from infection, understanding the cellular composition of the BAL could potentially impact treatment decisions for early BOS and provide critical data for developing diagnostic and prognostic biomarkers.

Surgical lung biopsy is no longer routine for posttransplant lung disease in the era of improved noninvasive diagnostics and empiric therapy but may be necessary to ascertain alternative noninfectious diagnoses when noninvasive means have not yielded definitive diagnoses (199, 200). The role of transbronchial lung biopsy for diagnosis of early BOS has not been definitively established. Transbronchial biopsy is less commonly performed in the HCT population than in the lung transplant population because of late symptomatic presentation, morbidity from cytopenias, associated complications, and low diagnostic yield for noninfectious disease entities (201–203). The performance and safety of transbronchial cryobiopsies in this context need to be assessed (204, 205).

Biomarkers

The U.S. Food and Drug Administration offers guidance on biomarker discovery and defines the following relevant biomarkers. First, a risk/susceptibility biomarker is an assay that indicates the potential for developing the disease in individuals who do not have clinically apparent disease. Second, a diagnostic biomarker is an assay used to confirm the presence of the disease. Finally, a prognostic biomarker is an assay used to identify likelihood of a clinical event or disease recurrence or progression in patients who have the disease (206).

Serum biomarkers that directly estimate BOS risk may circumvent some of the pitfalls of clinical risk factors, and clinical implementation of risk and diagnostic biomarkers for cGVHD are on the horizon (207). Among the few HCT BOS–specific biomarker investigations, serum MMP-9 (matrix metalloproteinase-9) has been shown to be increased at the time of BOS onset and is higher in those with BOS than in those with other forms of cGVHD; this finding is corroborated by proteomic analyses in HCT and lung allograft recipients (208–211). MMP-3 has also been shown to be increased in BOS (208, 209). B-cell activating factor and CD19(+) CD21(low) B cells were found to be increased in patients with a new diagnosis of NIH-defined BOS, consistent with a role in humoral immunity for BOS pathogenesis, but these findings have not been subsequently validated as a biomarker for BOS, and their role in identifying early BOS remains to be determined (212). In a single-arm trial of patients with BOS who were treated with pirfenidone, protein biomarkers that were associated with FEV₁ improvement included cathepsin D, EGFR (epidermal growth factor receptor), KIR3DL1 (killer cell immunoglobulin-like receptor 3DL1), and TNFSF14 [TNF ligand superfamily member 14]; these may represent markers of modifiable disease (111). Levels of the chemokines CXCL9 and CXCL11 were higher in patients with mild or moderate disease in a recent single-arm trial of ruxolitinib for BOS (15). Consistent with findings in CLAD-BOS, levels of CC16 were reduced in the serum of individuals with HCT-BOS (97). In a prospectively followed pediatric cohort, the glycoprotein KL-6 (Krebs Von Den Lungen-6) was found to be elevated early post-transplant in children who eventually developed BOS, compared to those who did not (213). Table 3 summarizes published HCT-BOS biomarker studies.

Table 3.

Serum and Plasma Biomarkers Identified in BOS after HCT

Biomarker	Association and Cohort	Subjects with HCT-BOS (n)	References
MMP-9	Increased in HCT-BOS	33	Inamoto et al. (208)
MMP-3	Increased in HCT-BOS and CLAD	88 (12 pooled, 76 confirmation)	Liu et al. (209)
BAFF, CD19(+) CD21(low) B cells	Increased, new diagnosis	46	Kuzmina et al. (212)
Cathepsin D, EGFR, KIR3DL1, and TNFSF14	Associated with FEV1 improvement in single-arm trial of pirfenidone	22	Matthaiou et al. (111)
CXCL9, CXCL11	Higher in mild/moderate disease in single-arm trial of ruxolitinib and BOS	49	DeFilipp et al. (15)
CC16	Decreased serum levels in HCT-BOS	8	Mattsson et al. (97)
KL-6	Higher serum levels at 1 month post-transplant in BOS compared with non-BOS; pediatric cohort	6	Gassas et al. (213)

Definition of abbreviations: BAFF = B-cell activating factor; CC16 = clara cell secretory protein; CLAD = chronic lung allograft dysfunction; CXCL9 = CXC motif chemokine ligand 9; CXCL11 = CXC motif chemokine ligand 11; EGFR = epidermal growth factor receptor; HCT-BOS = bronchiolitis obliterans syndrome after hematopoietic cell transplant; KIR3DL1 = killer cell immunoglobulin like receptor 3DL1; KL-6 = Krebs Von Den Lungen-6 glycoprotein; MMP = matrix metalloproteinase; TNFSF14 = tumor necrosis factor ligand superfamily member 14.

Emerging evidence regarding molecular markers associated with increased CLAD risk in lung transplant recipients may provide helpful insights for future biomarker studies aimed at risk prediction in HCT-BOS (97, 214, 215). Select circulating inflammatory measures may be a useful strategy to identify HCT recipients at greatest risk of future lung function decline. As RVI and many other CLAD risk factors reflect clinical events that potentially injure the airway epithelium, measures reflective of epithelial injury early after lung transplant may predict the subsequent PFT impairment such as specific small airway transcriptome inflammatory gene signatures on small-airway brushing sampling or low CCSP levels in BAL fluid (216).

Summary and key questions

A critical gap in our current strategy is identifying a sensitive but specific screening strategy; standard laboratory PFT alone does not reliably identify early disease, even when implemented at recommended screening intervals. Alternative modalities, including qCT, to assess incipient airway disease are being investigated, but none have been formally validated for early detection of BOS. A further limitation is the lack of an evidence-based longitudinal spirometric change metric in the current BOS definition. Predictive and diagnostic biomarkers, either circulating or at the tissue level, are also notably lacking (Table 1).

Research Road Map

We propose a framework centered around a multicenter cohort study and registry that will provide data to address the key research questions from major themes (Figure 2).

• 1.Conduct a multicenter prospective longitudinal observational study of allogeneic HCT recipients before lung disease. This will capture early events in the development of BOS; pre-, peri- and posttransplant risk factors; as well as physiology and biomarkers of early disease. Observation should start before the transplant through a minimum of 3 years of follow-up with protocolized clinical assessments and sample collection at landmark posttransplant time points. This study should leverage clinical implementation of PFT screening recommendations and incorporate newer early detection modalities such as home spirometry and qCT. Multicenter involvement will build on the single-center cohort study conducted by Bergeron and colleagues (1). The establishment of a longitudinal biospecimen repository including banked samples from clinically indicated bronchoscopy will extend the impact of existing biorepositories (including but not limited to those with the clinicaltrials.gov identifiers NCT05250037 and NCT05866302). Pairing these samples with prospective observational data that include the onset of infections and other key events will elucidate early events and help identify specific criteria for diagnosis earlier in the disease course.
• 2.Concurrently, establish a multicenter registry of incident and prevalent cases of HCT-BOS as part of the observational cohort. A registry with standardized collation of retrospective and prospective data will facilitate a better understanding of disease evolution and phenotypes and provide necessary data to update diagnostic definitions. Clinicopathologic correlations should be done with existing PFT, chest imaging, and lung biopsy specimens. Data can be mined using unbiased technology to find new clusters of patients with or without differences in outcomes. The inclusion of longitudinal clinical data will offer insights into treatment impact and allow us to align BOS outcomes with current and emerging BOS therapies, GVHD prophylaxis, and contemporary HCT practices. Particular attention should be paid to standardizing data collection and growth trajectories for children, who have the potential to recover lung function (217).

Figure 2.

Proposed research road map for early detection of bronchiolitis obliterans syndrome (BOS) after hematopoietic cell transplant. The research framework arising from a multicenter prospective longitudinal observational cohort that incorporates clinical screening and early detection practices, starting from pre–hematopoietic cell transplant through BOS diagnosis and beyond. Protocolized sample collection includes BAL fluid from clinical bronchoscopy and paired blood and nasal samples for multi-omics translational biomarker discovery. A patient registry of BOS will collate retrospective and prospective data. Specific knowledge gaps and key questions will be answered by this study design, defining phenotypes and diagnostic criteria, risk factors, and biomarkers. Concurrent preclinical investigations with novel in vitro techniques are also proposed. cGVHD = chronic graft-versus-host disease HCT = hematopoietic cell transplant.

We recommend the following high-priority clinical and translational investigations arising from these cohorts:

• 1.Revise the NIH diagnostic criteria to reflect the longitudinal nature of subacute lung function decline. Instead of a fixed threshold, an evidence-based spirometric change metric and other physiologic parameters should be emphasized (167, 174). Criteria should also take into account change metrics for individuals who have preexisting lung disease, such as COPD or asthma, because these are higher risk categories in which a change threshold may be different than in those without preexisting lung disease. Revision of the physiologic definitions would encourage the implementation of routine interval PFTs as a necessary screening measure and provide data to clarify the utility of specific spirometric parameters such as FEF_25–75, novel indices of obstruction such as FEV₃/FVC ratio, and associated changes in lung volume parameters such as residual volume/TLC ratio (174). An evidence-based refinement of BOS diagnostic criteria will harmonize definitions, improve diagnostic sensitivity for early detection, and define early disease endpoints for prospective studies and clinical trials. Criteria should be validated by nonoverlapping datasets and updated as needed with improvements in diagnostic and prognostic markers.
• 2.Investigate current barriers to the implementation of established screening and early-diagnosis clinical practice recommendations, particularly the use of laboratory PFTs. Conduct a survey of transplant centers to understand provider and patient attitudes and knowledge, accessibility of PFT laboratories, economic barriers, and patient-centered factors. A comparison of outcomes between centers that perform routine implementation compared with those that do not would provide insight into the impact of screening. Subsequent treatment decisions and FEV₁ trajectories should be correlated with meaningful longitudinal outcomes including survival, patient-reported outcomes, and quality of life. If early detection is effectively implemented, there should be a “stage shift” at diagnosis compared with historical published data or those from large HCT databases.
• 3.Discover and validate biomarkers for early detection. Establish risk/susceptibility biomarkers at landmark HCT events, including pretransplant, early posttransplant, and 1 year to answer key questions regarding the role of early lung injury and associated mechanisms. For early diagnosis, search for biomarkers in plasma and peripheral blood mononuclear cells with multi-omics approaches (e.g., proteomics, transcriptomics). Biomarkers should be investigated at the time of suspicion for BOS, e.g., early FEV₁ decrease. Physiologic biomarkers derived from laboratory PFTs and emerging home spirometry, as well as imaging biomarkers such as qCT strain metrics that can diagnose early disease, should be validated. From these data, a clinical algorithm for early detection that incorporates clinical risk factors with novel biomarkers should be established, which may help risk-stratify patients to determine who requires routine versus intensive screening approaches and when to initiate therapy for early disease.
• 4.Even though BAL analysis provides a limited picture of airway and parenchymal biology, this is the most tractable way to sample the local lung environment in patients. Bronchoscopy should be performed at the initial suspicion of disease. Conventional and -omics analyses of BAL fluid can answer key questions pertaining to the role of concomitant infections, immune cell populations, the lung microbiome, lung-specific biomarkers, and key drivers of pathogenesis. Non-lung samples obtained at the time of clinical bronchoscopy should include blood and nasal samples as well as stool samples to investigate the gut and lung microbiome interaction.
• 5.Explore and validate novel tissue sampling techniques that can identify early markers of disease. This can include minimally invasive sampling of nasal or bronchial epithelial lining fluid with an absorbent matrix that avoids dilutional effects with a high concentration of mediators and cytokines for more precise measurements (218). Markers of inflammation were readily detected in the nares in patients after HCT with “transient impairment.” These techniques could potentially circumvent the need to perform invasive bronchoscopies to distinguish transient impairment from progressive BOS (219).
• 6.Investigate novel interpretive and technical methods for identifying early lung dysfunction. Using existing data, derive PFT interpretive strategies that capture early changes in small airways that reliably predict progression to BOS. Use and study alternative physiologic assessments, including nitrogen breath washout, oscillometry, and radiomics to capture more sensitive information about functional and morphologic airways abnormalities. With a prospective observational study, determine the optimal timing of these tests for earlier disease identification and whether non-FEV₁ parameters might identify BOS.
• 7.Leverage ML and artificial intelligence on quantitative imaging and other diagnostic tests (e.g., spirometry) to explore a multimodality approach to early diagnosis that integrates clinical, physiologic, and circulating biomarkers. Newer artificial intelligence methodologies that can analyze small data sets should be explored (220). Unbiased clustering analysis that incorporates radiomics and lung function as in the case of asthma can also be applied to the multicenter patient cohorts (221).

Considerations for Risk Factor Analyses

Given the low incidence of BOS and the limited predictive performance of current screening approaches, identifying patients at the highest risk is crucial for targeted screening and early intervention (“prognostic enrichment”) (135, 169, 170). We recommend the following methodological considerations:

• 1.Sufficient patient numbers are necessary for adequate statistical power to test associations and to minimize the influence of practice variation including the heterogeneity among transplant recipients (age, indication, prior lung injury, prior antibiotic exposure) and transplant characteristics (preparative regimen, donor source, match, prophylaxis for GVHD).
• 2.Prospective studies of BOS risk factors require careful consideration of cohort selection. Although enriching for patients at high risk can improve study feasibility, overly restrictive enrollment criteria may introduce selection bias and miss key windows of disease onset. We recommend a two-stage approach: first, identify preliminary risk factors (accounting for patient, transplant, and post-HCT variables) with a large prospective cohort study with broad inclusion criteria, followed by validation of these findings in focused case-control studies using independent patient populations.
• 3.Risk factors may change as the treatment landscape evolves, and therefore studies should enroll patients briskly enough that major changes in cGVHD risk do not influence study outcomes and conclusions. Updated diagnostic criteria for BOS that encompass a preclinical disease definition should be employed.
• 4.Causal inference studies should focus on purposeful variable selection and work within a framework of prespecified causal models using directed acyclic graphs (144). Models that examine risk factors that occur with variable timing after HCT (such as occurrence of RVI) should explore a more dynamic strategy (e.g., extended Cox proportional hazard models) with a careful consideration of how long a risk factor (e.g., RVI or extrapulmonary cGVHD) will continue to influence the subsequent risk for BOS (144). Nondichotomous variables, including continuous PFT parameters, as well as a combination of multiple cross-sectional and longitudinal factors, should be considered for the development of risk-assessment tools.

Pathogenesis and Early Detection

We recommend the following concurrent preclinical investigations of early disease pathogenesis, which will inform the translational work on human samples and vice versa.

• 1.Develop and use in vivo and in vitro models to test the hypothesis that antecedent lung injury mediated by alloimmunity, infection, drug toxicity, aspiration, or a combination of insults contributes to BOS development. Air–liquid interface culture of airway epithelium or organoids that can approximate airway structures could be an alternative to live animal models, mitigating the problem of disparate airway anatomy across species, as these can be generated from human lung cells (222–224). These three-dimensional structures could be used to evaluate the role of particular immune or lung cells and incorporate the endothelium.
• 2.Focus on the role of the lung epithelium in preclinical models. Specifically, investigate the interactions of progenitor cells (e.g., CCs, basal cells) with immune cells including macrophages and fibroblasts. Subsequent translational confirmation using ex vivo approaches such as xenografts to investigate CCs and basal cells from HCT recipients with early disease will also be key (225). Analysis of human biopsies with spatial transcriptomics or mass imaging could help to better describe the cross-talk between immune and resident cells in the lung tissue (226). Bronchial epithelial brushings or endobronchial biopsies may facilitate primary human cell culture models to understand which perturbations in epithelial cells lead to BOS, such as has been investigated in asthma and other airway diseases (227, 228).
• 3.Use the enhanced understanding of disease pathogenesis to inform timing of early detection and therapeutic targets. Given that the disease likely involves multiple insults and stages of immune dysregulation and progression to airway fibrosis, capturing the timing of this progression could identify the optimal therapies or combination of therapies to implement preemptively, at initial diagnosis, and at different stages of disease progression.

Additional Gaps and Future Directions

In addition to scientific knowledge gaps, the committee identified additional clinical care gaps and future directions pertaining to the early detection of BOS.

Pulmonary–HCT Collaboration

There is a need for greater interaction between pulmonary and transplant specialists to improve early diagnosis and provide guidance on diagnostic quandaries when disease is suspected. Similarly, collaboration among subspecialist pulmonologists, community pulmonologists, and transplant providers is critical and relevant to patient care because most HCT survivors return to the community for longitudinal care rather than stay at the tertiary HCT center. Therefore, it is imperative that clinicians—hematologists, pulmonologists, and general practitioners—possess basic knowledge about BOS and understand the rationale behind early detection. A key role of the pulmonologist is to collaborate with HCT physicians to facilitate early detection by removing barriers to obtaining PFT results and to disseminate the evidence base to practitioners and patients alike. We recommend that HCT centers develop teams with specific pulmonology expertise to manage patients with BOS and other LONIPCs (229).

A Distinct Clinical Entity

Barriers to research, as well as clinical care, include lack of recognition of BOS and other LONIPCs as distinct clinical entities in diagnostic codes, professional societies, and funding agencies. HCT-BOS is often categorized with lung transplantation or obstructive lung disease and occasionally among rare diseases. This impedes the efforts of clinicians and researchers to create an appropriate infrastructure for a patient-accessible and comprehensive multicenter registry, as has been done for other rare lung diseases such as cystic fibrosis, to understand the disease and to conduct clinical trials.

Interventional Trials

The impact of early detection depends on whether it is actionable with targeted therapies. Systemic evaluation of the impact should occur in the context of a well-designed and controlled clinical trial that incorporates early detection with therapy targeting lung inflammation as well as fibrosis. Study design, including the timing and duration of intervention and choice of pharmacologic agent, should be guided by lung-specific early disease biomarkers and knowledge of disease evolution (230). This should include randomized trials to differentiate the effect of the drug from the natural history of the disease. Clinical endpoints depend on the phase and phenotype of the disease; if early disease is characterized predominantly by inflammation, and the therapy is directed toward inflammatory pathways (e.g., ruxolitinib), the benchmark for efficacy should be improvement or resolution of incipient lung dysfunction within a short time frame, e.g., 1 month. The goal of upfront therapy with agents targeted toward fibrosis should result in stabilization of lung function decline in the short term and improvement in lung function trajectory in the long term, as suggested by a recent trial with pirfenidone (112). To date, only a handful of clinical trials specifically for HCT-BOS have been completed, and all of them tested agents developed or approved for other indications (7, 15, 111, 195, 231, 232). Rationally designed clinical trials based on updated knowledge of the natural history of HCT-BOS should be considered a research priority for the near future.

Conclusions

Significant knowledge gaps exist for BOS after allogeneic HCT, which impede early detection. Increasing awareness of BOS is necessary because more HCT survivors return to community care. We propose a multipronged research approach to address these gaps. The collaboration of a critical mass of pulmonologists, hematologists, and investigators from allied disciplines is necessary to conduct this research and to improve outcomes.