Approaching Clinical Trials in Childhood Interstitial Lung Disease and Pediatric Pulmonary Fibrosis

Abstract

Childhood interstitial lung disease (chILD) comprises a spectrum of rare diffuse lung disorders. chILD is heterogeneous in origin, with different disease manifestations occurring in the context of ongoing lung development. The large number of disorders in chILD, in combination with the rarity of each diagnosis, has hampered scientific and clinical progress within the field. Epidemiologic and natural history data are limited. The prognosis varies depending on the etiology, with some forms progressing to lung transplant or death. There are limited treatment options for patients with chILD. Although U.S. Food and Drug Administration–approved treatments are now available for adult patients with idiopathic pulmonary fibrosis, no clinical trials have been conducted in a pediatric population using agents designed to treat lung fibrosis. This review will focus on progressive chILD disorders and on the urgent need for meaningful objective outcome measures to define, detect, and monitor fibrosis in children.

Childhood Interstitial Lung Disease/Diffuse Lung Disease: An Overview

Childhood interstitial lung disease (chILD) comprises a complex and heterogeneous spectrum of >200 rare respiratory disorders (immune mediated and non–immune mediated) that affect infants and children (1). The pathological mechanisms differ, with distinctly different types of disorders presenting in children <2 years of age compared with those 2–18 years of age (2, 3). The reported incidence and prevalence of chILD vary widely (4–9), and the relative frequency and type of ILD differ between children and adults (10, 11), with chILD being comparatively underrepresented in pulmonary research (12). Even when disorders overlap with adult ILD, there is a potential for different disease manifestations and trajectories within the developing lung, requiring tailored, child-centered approaches to clinical management (1, 8, 13, 14). Reported mortality rates vary among centers (3, 4, 9, 15–17) and a sampling effect for higher mortality rates has been observed with the use of lung biopsy ascertainment (3), likely related to sampling bias, as patients who require a biopsy typically have more severe disease. The severity of disease at initial patient evaluation was found to correlate with prognosis, potentially serving as a measure of outcome (18).

Morbidity and mortality rates vary depending on the specific chILD disorder. For example, infants with neuroendocrine cell hyperplasia of infancy typically improve with age, with no apparent impact of the disease on mortality (19), whereas infants with mutations in the gene for ABCA3 (ATP-binding cassette transporter A3) have high rates of morbidity and mortality (20–22). Variable morbidity and mortality rates have also been observed in antineutrophil cytoplasmic antibody–associated vasculitis, connective tissue diseases (e.g., systemic lupus erythematosus, systemic sclerosis, and Sjögren’s syndrome), dermatomyositis, juvenile idiopathic arthritis (JIA) (e.g., systemic JIA and rheumatoid factor–positive–positive JIA), sarcoidosis, COPA (coatomer protein complex subunit α) syndrome (a genetic disease with an autoimmune and autoinflammatory profile), and STING-associated vasculopathy with onset in infancy (SAVI), which is associated with TMEM173 mutations. Moreover, the field is continually being redefined by the increasing identification of genetic mechanisms, particularly in surfactant dysfunction disorders and genetically driven immune disease (23). Clinical presentation and mortality rates vary based on the mutation type (24), and outcomes vary by age at presentation (21).

To date, no clinical trials have been conducted in children using agents designed to treat lung fibrosis, despite the availability of U.S. Food and Drug Administration (FDA)-approved treatments for adult patients with idiopathic pulmonary fibrosis (IPF) (25, 26). In addition to the barriers of diagnosis and identification, our understanding of chILD is limited by the lack of defining fibrotic phenotypes in children and by lung disease occurring on the background of lung development, with disease progression not necessarily proceeding in the manner described in adults. Critical consideration is also required to include the formulations of drug therapy, safety assessment, and optimizing outcome measures that are reliable and predictable for this population. This review will explore the issues associated with chILD in terms of diagnosis, the pathophysiology of fibrosis, the currently available treatment options for patients with chILD, and the challenges associated with investigating chILD in clinical trials, with recommendations for research and treatment advancement.

Diagnosis of chILD

A diagnosis of chILD is based on variable definitions, including multiple disorders, some of which are poorly understood (2, 8, 16, 27), and limited epidemiologic (5) and systematic clinical (2) data. chILD syndrome is usually diagnosed if three of the following features are present: 1) respiratory symptoms (cough, rapid and/or difficult breathing, and exercise intolerance), 2) respiratory signs (tachypnea, adventitious sounds, retractions, digital clubbing, and failure to thrive or respiratory failure), 3) hypoxemia, and 4) diffuse abnormalities on a chest radiograph or computed tomography (CT) scan (2, 3).

The choice of which specific diagnostic test to use depends on a variety of factors, including the clinical context and disease severity and duration (2). Clinical evaluation includes chest radiography, chest CT, pulmonary function testing, infant pulmonary function testing, bronchoscopy with BAL, echocardiography, genetic testing, and/or lung biopsy (2).

A diagnosis is typically made based on the phenotypic pattern resulting from the patient’s history, physical exam, and evaluation findings. In a number of circumstances, lung biopsy may not be required for a diagnosis of some forms of chILD. For example, CT scans can identify specific imaging patterns for diagnoses (Figure 1). However, radiologic–pathologic correlations have not been well studied for most types of chILD (28). American Thoracic Society guidelines recommend a stepwise approach using less invasive diagnostic modalities. A historical retrospective review revealed that 68–75% of patients who had received a diagnosis had undergone a lung biopsy (9, 15). If a lung biopsy is required, a single-center study found that they were well tolerated in infants with suspected ILD (29). Several studies (16, 30–32) and our collective experience have demonstrated that lung biopsies, with input from experienced pathologists, are beneficial in the diagnosis and management of idiopathic diffuse lung disease in children. In patients with chILD undergoing a surgical lung biopsy, the American Thoracic Society Clinical Practice Guidelines recommend video-assisted thoracoscopy, rather than open thoracotomy (2, 33), and a standardized approach to processing (34).

Figure 1.

Chest computed tomography (CT) findings in patients with childhood interstitial lung disease. (A) Axial CT image from a 10-year-old girl with systemic sclerosis, showing minimal areas of ground-glass attenuation, septal thickening, and peripheral honeycombing. (B) Axial CT images from a patient with juvenile systemic sclerosis. The left panel shows reticular and ground-glass opacities at the periphery when the patient was 12 years of age; the right panel shows that clusters of cysts have developed along the periphery in the same patient at 18 years of age. (C) Axial CT image from a 16-year-old girl with systemic lupus erythematosus, showing large areas of ground-glass attenuation superimposed on interlobular septal thickening (“crazy paving”). The right panel shows a high-resolution CT image from a 15-year-old girl with systemic lupus erythematosus and lupus pneumonitis showing a similar pattern. (D) CT scans from an 18-year-old with a genetic SPC (C460 + 1 G→A) showing reticular opacity, honeycombing, traction bronchiectasis, architectural distortion, and cystic opacity (without walls). This is not a typical adult pulmonary fibrosis pattern. A and C are reprinted from Reference 81 by permission from the Radiological Society of North America. B is reprinted from Reference 82 by permission from Elsevier.

Genetic testing has also allowed for a noninvasive approach to diagnosis (15, 35). Genetic abnormalities that should be evaluated when a diagnosis of chILD is suspected include deletions or mutations in genes involved in surfactant production and/or function (SFTPB, SFTPC, ABCA3, and NKX2.1), surfactant catabolism (CSF2RA and CSF2RB), and lung development (Fox F1, FLNA, TBX4, and FGFR2, among others) (22, 35, 36) depending on the clinical context. Genetic testing and blood markers of rheumatologic disease may also be diagnostic without biopsy. Patients with neuroendocrine cell hyperplasia of infancy have shown a genetic basis that has been identified in a minority of cases to date (15, 16, 37). Although obviating the need for lung biopsy is advantageous for the patient, fewer lung biopsies are now being performed for chILD, which may limit our understanding of its histopathologic features and assessment of fibrosis.

Pathophysiology of Fibrosis in Specific Forms of chILD

chILD diseases with significant pulmonary fibrosis are well recognized (Table 1) (3), although not all forms of chILD are associated with pulmonary fibrosis. The impact of chILD-related fibrosis on the developmental trajectory of the lung is also poorly understood, though there may be overlapping pathways. For example, transforming growth factor-β, which is a critical regulator that governs the induction of fibrosis, including in diseases such as systemic sclerosis, also plays key roles within the context of lung growth (38, 39). Children’s lungs continue to grow and develop well beyond infancy, a process that involves alveolar acquisition and increased complexity, alteration of the size and length of the airways and vasculature, and a change in cell population. Ongoing postnatal lung development can impact disease processes and has implications for therapeutic interventions.

Table 1.

Childhood Interstitial Lung Disease Conditions Potentially Associated with Pulmonary Fibrosis (3, 79, 80)

Immunocompetent host (hypersensitivity pneumonitis or bronchiolitis obliterans)

Surfactant disorders

SFTPC mutations

ABCA3 mutations

NKX2.1 gene mutations

Immunocompromised host

Opportunistic infections related to treatment (chemotherapy, radiation, or drug hypersensitivity), transplantation and rejection, or lymphoid infiltrates

Radiation pneumonitis

Posthematopoietic stem cell transplantation

Disorders related to systemic disease processes

ANCA-associated vasculitis

Collagen-vascular disease

Genetically driven diseases such as COPA and SAVI

JIA: sJIA or RF-positive poly JIA

Langerhans cell histiocytosis

Mixed connective tissue disease

Nonspecific interstitial pneumonia

Sarcoidosis

Systemic lupus erythematosus

Storage diseases

Systemic sclerosis

Definition of abbreviations: ABCA3 = ATP-binding cassette transporter A3; ANCA = antineutrophil cytoplasmic antibodies; COPA = coatomer protein complex subunit α; JIA = juvenile idiopathic arthritis; NKX2.1 = NK2 homeobox 1; RF = rheumatoid factor; SAVI = STING-associated vasculopathy with onset in infancy; SFTPC = surfactant protein C; sJIA = systemic juvenile idiopathic arthritis.

Causal mechanisms of fibrosis are likely to include a genetic component but may also include microbial triggers, inhalational injury, and autoimmune dysregulation (40, 41) (Figure 2). It is unknown whether mechanisms that govern fibrosis in the context of the aging lung also apply to children when lung development is ongoing. One important difference between fibrosis in children and that in adults is the lack of fibroblastic foci in the former. Except in very rare cases, children do not get usual interstitial pneumonia (42). As demonstrated in lung pathology, chILD fibrosis often involves a mixed inflammatory and fibrotic process, as well as surfactant disruption.

Figure 2.

Epithelial injury hypothesis of idiopathic pulmonary fibrosis. Reprinted from Reference 83.

A “progressive fibrosing” (PF) phenotype has been identified in adult ILD, characterized by an initial fibrotic response to lung injury that becomes progressive, self-sustaining, and independent of the underlying etiopathology or injurious agent/trigger (43). Within the context of chILD fibrosis, the concept of a PF phenotype has yet to be addressed or accepted by specialists. Data from adult patients with ILD and a PF phenotype may help us understand how to define a similar phenotype in chILD. On imaging and pathologic assessment, an apparent PF phenotype has been observed in conditions such as genetic surfactant disorders (e.g., mutations in SFTPC [44] and ABCA3 [21, 45]), rheumatologic disease (41, 46), and radiation-induced injury (47); however, no detailed natural history studies have been performed. The classic radiologic findings observed in adult patients with IPF, such as honeycombing and traction bronchiectasis, are often not present. In chILD disorders with fibrosis, chest CT scans often show early prominent ground-glass opacification with progression to a cystic change with reticular abnormalities (48). The geographic distribution of fibrosis in chILD is also dissimilar to IPF, making it difficult for radiologists or pattern recognition tools to identify these rare progressive chILD cases.

It is challenging to define the terms “progressive” and “fibrosis” due to the lack of accepted criteria (49), and thus additional data are required. Adding to the complexity, the mechanisms that drive disease progression in chILD are likely affected by the initiating etiopathologic event and the host environment (40). Disrupted epithelial plasticity and perturbed epithelial–mesenchymal cross-talk, which are modulated by genes that govern lung development and embryonic signaling pathways, may play a role (40). Therapeutic targeting of the PF phenotype using antifibrotic agents to abate disease progression in adults is currently undergoing evaluation in phase II (ClinicalTrials.gov identifier: NCT03099187) (50) and phase III (ClinicalTrials.gov identifier: NCT02597933 [SENSCIS (Safety and Efficacy of Nintedanib in Systemic Sclerosis)] and ClinicalTrials.gov identifier: NCT02999178 [INBUILD]) randomized controlled trials (43).

Currently Available Treatments for chILD

There is currently an urgent need to develop opportunities for FDA-approved medications and new drug therapies for chILD. Current clinical management guidelines focused on diffuse lung disease largely rely on open-label or retrospective studies, or data from individual cases (41). Treatment options include oxygen, corticosteroids, hydroxychloroquine (HCQ), and azithromycin (4, 9, 17, 51, 52), and are based on the concept of suppressing inflammation to prevent progression to fibrosis (4, 17, 53). When chILD is associated with a rheumatologic diagnosis, guidelines based on that systemic diagnosis are used for treatment. Supportive care includes annual routine immunization, treatment of infection, avoidance of air pollutants, and invasive and noninvasive ventilation when required (53). New treatment options may be forthcoming with a better understanding of the disease process in animal models of disease (54) (Figure 3) and the potential for high-throughput drug testing in cell cultures developed from inducible pluripotent stem cells for the surfactant dysfunction mutations (55). However, improved infrastructure and strategies are needed to translate these discoveries to patients. A limited number of small, multicenter clinical trials in chILD have focused on nonspecific treatments for fibrosis, such as glucocorticoids and HCQ, with unclear results (4, 9, 17, 52, 53). Currently, there is one open trial in chILD in Europe, which is a phase IIa, randomized, double-blind, placebo-controlled investigation of the initiation or withdrawal of HCQ in pediatric ILD (ClinicalTrials.gov identifier: NCT02615938).

Figure 3.

(A) Mechanism of lung inflammation and remodeling in an SFTPC (surfactant protein C) BRICHOS mutant mouse model with cytokines from childhood interstitial lung disease (chILD) (54). A critical substitution at C121G (cysteine-to-glycine substitution at codon 121) in the SFTPC gene identified in a patient with chILD, when expressed in adult mice, resulted in an endoplasmic reticulum (ER)-retained proprotein and alveolar type 2 (AT2) cell ER stress. AT2-cell–derived cytokines in the mouse model were identified in BAL fluid (BALF) supernatants from patients with chILD, and STPF^C121G mice developed pulmonary fibrosis (54). (B and C) Histopathologic figure of lung specimens from two patients with SFTPC mutations at (B) 21 months of age, demonstrating chronic pneumonitis of infancy, and (C) 21 years of age with advanced fibrotic disease. ^aNon-AT2 source. ^bComp. UPR = unfolded protein response.

Challenges of Investigating chILD in Clinical Trials

There are currently no Cochrane Systematic Reviews available for chILD or pediatric pulmonary fibrosis, and patient management relies mainly on case reports, anecdotal evidence, and ad hoc observations (2). Moreover, low patient numbers have made it difficult to achieve an adequate sample size for clinical trials. To address this issue, treatment protocols and a Delphi consensus process were created in an attempt to harmonize clinical approaches (27). Collaborative efforts to gain insight into chILD include the Rare Lung Disease Consortium (56), the Children’s Interstitial and Diffuse Lung Disease Research Network (chILDRN) (57), the Children’s Interstitial Lung Disease Foundation (58), and LungMAP (59, 60). A consensus has been reached on the need to connect international registries to act as curated repositories of standardized, systematic epidemiologic and clinical data of sufficient quantity and quality to enhance our understanding of chILD and improve clinical management (2, 12). Existing databases include the chILD-EU Register, RespiRare (French national register), Australasian Registry Network for Orphan Lung Disease (ARNOLD), and chILDRN. In addition to registries and collaboration among international investigators, adaptive trial designs may help circumvent the requirement for a large group of patients (61). To facilitate the clinical trials necessary to advance medical understanding and improve patient outcomes in chILD, several key challenges must be addressed (2).

Diagnostic and monitoring techniques, including the identification of noninvasive biomarkers of lung structure, function, development, and remodeling, should be improved and standardized. Technical limitations hamper attempts to define fibrosis or diagnose it in pediatric patients, due to differences in chest CT patterns and prognosis compared with adult fibrotic disease. Trial endpoint measures that are appropriate for evaluating fibrotic disease in the developing pediatric lung must be identified and validated. Lung histopathology represents the gold standard for diagnosing fibrosis, but in many cases, current genetic testing–based diagnosis in clinical practice has resulted in fewer lung biopsies. In addition, when biopsies are obtained, it is often during the early stages of disease rather than later, when fibrosis has fully developed. Finally, there may be pharmacologic differences between adults and children that affect the preparation and pharmacokinetics of treatments that were developed for older patients.

Biomarkers on the Background of a Developing Lung

Lung development is an ongoing process in childhood, and lung function may increase in the growing lung, even in the presence of an inflammatory process. In adult patients, therapies that mitigate but do not stop this process would merely slow the rate of lung function decline; however, in children, lung function may improve because of continuing lung growth, rather than reduced inflammation. This novelty illustrates that adult clinical trial results may not be generalizable to pediatric studies. Pulmonary function measurements are also limited by whether the patient is old enough to reliably complete tests accurately, although modalities are increasingly being developed for application in preschool-age children, including infant pulmonary function tests (infants), lung clearance index, and oscillometry (preschool children).

Beyond lung function, few biomarkers for pulmonary fibrosis in children have been identified. Biomarkers in children with SFTPC mutations were shown to be similar to an animal model of spontaneous pulmonary fibrosis using an SFTPC mouse model (Figure 3) (54). Some of the same biomarkers, such as MMP-7, IL-6, and IL-8 (54, 62, 63), have been reported in adult IPF, making a compelling argument for evaluating these markers of disease in children (64).

Identifying a core outcome set may provide an efficient way to compare results between different clinical trials (65); however, a core outcome set is not currently available for chILD due to the challenge of defining consistent clinical outcome measures for children. Although discussions with the regulatory authorities may allow further exploration of this topic, there is an unmet need for additional, novel outcome measures, including physiologic and revised infant pulmonary function measurements. It may be easier to evaluate measurable changes in exercise, respiratory rate, and pulse oximetry in this patient group, which may lead to improvements. In addition, the impact of treatment on patients and their families is also important and should be captured via quality-of-life assessments. Quality-of-life instruments have been developed to quantify impairments in health-related quality of life in patients with chILD (66, 67).

Challenges in Defining and Monitoring Fibrosis on Chest CT

A large number of CT scans collected from patients with IPF from the Idiopathic Pulmonary Fibrosis Clinical Research Network have demonstrated the clinical value of CT in rare diseases such as IPF (68). The Idiopathic Pulmonary Fibrosis Clinical Research Network has conducted organized, high-quality research studies, and has shown that CT findings of fibrosis have a high prognostic value, i.e., CT findings are highly correlated with mortality (69, 70). CT patterns in IPF and non-IPF are also well documented, making chest CT a valuable diagnostic marker in older patients. Although CT analysis will be an integral tool in chILD clinical trials, significant differences between pediatric fibrotic lung disease and IPF, in both imaging patterns and disease outcomes, make existing imaging criteria not easily transferable.

To be a useful outcome marker and clinical tool, chest CT definitions must be standardized in ILD and findings must be scored for use in clinical trials (e.g., expert reader scores and/or computer-based methods for quantifying disease) (71–74). Validating these outcomes in a rare pediatric disease is difficult, so future global collaboration is essential to increase the number of patients evaluated using the same clinical tools.

Consideration should be given to the decreasing lung density seen with age/growth in childhood because this causes challenges with longitudinal comparisons, especially using quantitative CT tools. Changes with age mean that greater numbers of patients will be required to power a study with children of different ages appropriately. In adult IPF, automated texture analysis has shown a strong prognostic value (70, 75, 76). This approach may be feasible in children when quality chest CT technology is available. Validation of these quantitative CT outcomes will also require multisite global collaboration to enroll the necessary numbers of subjects.

Use of anesthesia with intubation and controlled ventilation, with appropriate inflation pressures to 25 cm H₂O preceded by recruitment maneuvers, is typically required to obtain good-quality inspiratory chest CTs in infants (77); however, improvements in nonsedated free-breathing chest CT may provide an option to limit the risks associated with anesthesia. In older children, full inspiratory breath-hold scans can be performed in addition to the spirometer-controlled chest CT imaging that is used in specialized centers. The pediatric chest radiology community should aim to reach an agreement on imaging strategies in chILD. Given that there are already differences between lung densities in adults and children, it is important to 1) standardize CT scanning across infants, preschoolers, and school-age children; and 2) consider the effect that tidal breathing chest CT imaging will have on infant chest CT imaging. The potential role of lung magnetic resonance imaging and other novel imaging outcomes to determine fibrotic features in children is not yet clear.

When describing radiologic assessments, radiologists should aim to use specific terms associated with fibrosis (e.g., honeycombing, traction bronchiectasis, and reticular abnormality) when appropriate in children. In addition, they should explain that these findings can be seen in fibrotic lung disease but are not indicative of or in the pattern of classic adult IPF. It is important to be clear that fibrotic findings on chest CT in children are not equivalent to a diagnosis of adult IPF. Correlation with any biopsy finding is required, as biopsies may show fibrosis with nonclassic fibrotic features on chest CT, including cysts and architectural distortion. Lung biopsy should be considered if a patient could potentially benefit from an antifibrotic drug and chest CT findings were nondiagnostic.

Safety

In IPF, morbidity and mortality are high, and the perceived risk of repeated radiation is outweighed by the benefit. In chILD, the disease may be initially milder and less severe, and repeated radiation exposure may pose a greater risk to the developing child. Fortunately, up-to-date chest CT protocols can provide acceptable image quality with ultra-low-dose imaging (0.2 mSv).

There are additional considerations in children compared with adults with regard to safety. Treatments may impact developing lungs differently than they would adult lungs. The effect of therapies on liver function is also likely to differ between children and adults (78). In addition, side effects may extend to other developmental processes, including the development of teeth, growth plates of the long bones, and the intestinal epithelial lining. This safety profile has yet to be evaluated.

Conclusions

chILD comprises more than 200 heterogeneous and rare disorders. Significant morbidity and mortality rates are seen in chILD, but poor definitions and identification have impeded improvements in treatment. In this time of antifibrotic drug trials in adults, the need to define fibrotic lung disease in children is urgent. In addition to patient histories and physical exams, good-quality, standardized chest CT; genetic testing; and emerging biomarkers are critical. Descriptions of the imaging patterns observed in chILD are required, especially when there are differences in fibrosis patterns and outcomes between pediatric and adult phenotypes. Moreover, scoring systems for pediatric fibrosis and other ILD features require development and validation in children. Evaluations of the safety of antifibrotic medications in the developing lung are needed, and to help patients with chILD, we must come together to achieve progress on treatment options. This includes collaboration in advancing our understanding of fibrosis in children, and identifying suitable outcome measures and biomarkers to facilitate clinical trials in chILD, particularly in patients with progressive fibrosis.