The starting point: when to suspect ChILD—typical presentations

Initial investigations

Given the non-specific nature of the presentation of chILD the probability of a diagnosis can be enhanced by structured investigation. A simple severity score has been proposed (table 1),20 which sadly has yet to be superseded by more sophisticated biomarkers:

1. Clinical: Evidence of hypoxaemia is often the first abnormality to raise concern and should be noted whether it is present at rest both awake and asleep breathing room air. In older children exercise testing should be used to unmask desaturation which is not present at rest, following standard protocols; there is insufficient evidence to recommend a particular method over others. It is unwise and uninformative to exercise a child who is desaturated at rest, without giving supplemental oxygen.

2. Radiology: The chest radiograph may be normal, but more likely reveals non-specific abnormalities. Gastro-oesophageal reflux is common in young children and also in patients with respiratory disease; deciding if reflux may be contributory is difficult even in the presence of a barium swallow reflux study and a pH study.

3. Physiological testing: Lung function testing in infants has little general availability and the role in infants is unclear. Neuroendocrine cell hyperplasia of infancy (NEHI) is characterised by a pattern of hyperinflation and airflow obstruction,21 but there is insufficient experience to rely on physiological patterns in infancy in chILD. In older children, the usual pattern is classically restrictive, with a reduced FEV₁ and FVC, a normal or even elevated FEV₁/FVC ratio and reduced lung volumes. Carbon monoxide transfer (diffusing capacity or transfer factor of the lung for carbon monoxide, DLCO) is usually reduced, and if elevated, pulmonary haemorrhage should be suspected.22 The occasional chILD may have a more obstructive pattern of physiology.

4. Echocardiogram: Should be an early investigation to estimate pulmonary artery pressure, and exclude cardiac mimics of interstitial lung disease, such as cor triatrium leading to pulmonary oedema.23 Pulmonary hypertension, where present, should be diagnosed and treated accordingly in liaison with paediatric cardiology colleagues.

5. Blood studies: A focused approach is recommended. Tests can be grouped into (a) genetic abnormalities; (b) immune function (especially if the follicular bronchiolitis-lymphoid interstitial pneumonia spectrum is suspected); (c) auto-antibody studies (cases of pulmonary haemorrhage, alveolar proteinosis, or if there is evidence of a systemic disease); (d) environmental organic dust exposures (hypersensitivity pneumonitis); (e) miscellaneous, that is, ACE inhibitors in cases of suspected sarcoidosis. Note that novel genetic chILD entities with multisystem disease are being described,24 ,25 and the younger the child is the more carefully new or established genetic diagnoses are sought. In all cases of chILD, DNA of the patient and parents should be stored for future analyses. The clinical situation will dictate which tests are performed, and whether it is realistic to await results before proceeding to a CT or lung biopsy.

A proposed diagnostic pathway is shown in figure 1. In terms of timing of enrolment of suspected chILD patients in the registry, a preliminary discussion at an early stage would be valuable for the patient. The use of, for example, standardised scanning protocols is to be preferred.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Proposed flow chart for investigating interstitial lung disease in children (chILD). If the child is well and stable, progress to the more invasive investigations may not be indicated. HRCT, high-resolution CT.

If chILD is suspected clinically: the role of chest CT scanning

CT chest is invariably performed in the investigation of a possible case of ChILD. The purpose of chest CT is to evaluate the presence and extent of disease. In some cases it can be diagnostic,26 but is more usually supportive of a diagnosis that takes account of clinical history, blood results and sometimes bronchoalveolar lavage and/or biopsy material.

CT scanning for chILD should only be performed in centres experienced in paediatric radiology. The general principles of scanning are to minimise radiation dosage to the child while maximising the information obtained; a very low dose scan which is diagnostically useless benefits no one.27 Detailed protocols for reducing radiation exposure while maintaining image quality are given online, and these should be audited regularly.

The optimal CT scan is a volumetric scan during inspiration, performed in tandem with a high-resolution CT (HRCT) fine-cut spaced expiratory scan. Ventilation should be controlled to ensure satisfactory, interpretable, scan output. In young children this necessitates a general anaesthetic (ie, under 5 years of age), although some centres advocate sedation and controlled, bag and mask ventilation.28 As always, the risk of anaesthesia or sedation should be weighed against the risk of not obtaining diagnostic information. Older children should be able to reliably hold breath in inspiration and expiration for 6–10 s. In children who are ventilated under a general anaesthetic it is important to attain appropriate ventilation pressures in both inspiration and expiration; these details are provided in our online standard operating procedure (SOP).

The administration of contrast medium will make the assessment of ground glass shadowing almost impossible, and so careful consideration should be given as to the risk/benefit of using a contrast medium based on the anticipated diagnosis. The need to assess the pulmonary vasculature will require the use of contrast, with detrimental effects on the interpretation of some lung components.

The three outcomes of HRCT scanning (figure 1) are (1) chILD unlikely including that the CT is normal. Although a normal CT scan is unusual in ChILD, it may occur in some cases early in the course of the disease process where there are very few non-diagnostic CT features; (2) chILD present and a specific diagnosis can be made26; and (3) chILD present, with or without possible evidence of a comorbidity such as aspiration (eg, consolidation in a typical distribution), but a specific diagnosis cannot be made. If chILD is unlikely then appropriate diagnostic algorithms to investigate possible non-chILD findings are deployed; these are beyond the scope of this article. A specific diagnosis of chILD may allow immediate action; however, the ability for CT to be completely diagnostic is limited to relatively few conditions (ie, pulmonary alveolar proteinosis, extrinsic allergic alveolitis and NEHI), and these too are usually supported by other diagnostic tests, that is, environmental allergen and prednisolone response in hypersensitivity pneumonitis (figure 2A) ,or granulocyte-macrophage colony-stimulating factorreceptor auto-antibody or gene mutation studies, surfactant protein gene mutations and an immunological work-up in cases of pulmonary alveolar proteinosis (figure 2B).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Diagnostic high-resolution CTs (HRCTs) in childhood interstitial lung disease (chILD). (A) There is diffuse ground glass shadowing with a widespread soft centrilobular nodular pattern. This is virtually diagnostic of hypersensitivity pneumonitis, in this case due to pet doves. (B) There is generalised ground glass opacification with thickening of the interlobular septa, giving the classical cobblestone appearance of pulmonary alveolar proteinosis. It is not possible to distinguish the different aetiologies from the HRCT appearances.

It is recognised that although faster CT scanners may enable chest CT without the need for anaesthetic, we do not recommend this unless anaesthesia is thought to be unsafe, as (1) the technique frequently provides suboptimal results from respiratory movements made by the child and (2) the variance in lung volumes during an uncontrolled respiratory cycle will reduce the value of the scans obtained and the reliability of CT to provide the supportive evidence for a diagnosis of ChILD.

Whatever the timing of investigations, planning to avoid the child having multiple general anaesthetics is essential. So fibre-optic bronchoscopy should only be done as a separate procedure if it is likely to be diagnostic; otherwise, it is better performed under the same anaesthetic as the lung biopsy.

Invasive testing: flexible bronchoscopy

Bronchoscopy with bronchoalveolar lavage (BAL) may be diagnostic in pulmonary haemorrhage syndromes, alveolar proteinosis and eosinophilic lung disease, and a normal cell differential can rule out hypersensitivity pneumonitis. BAL target site should be chosen on the basis of radiological abnormality or airway inspection. If BAL is to be performed during the same anaesthetic as CT, then it should follow the imaging; where it is to be performed at the time of lung biopsy, the lobe designated for a lung biopsy should be avoided. Where possible flexible bronchoscopy should be performed via endotracheal or laryngeal mask to reduce suction channel contamination from the upper airway. BAL volume is adjusted to body weight using 3 mL/kg of sterile normal saline divided into three equal fractions in children weighing <20 kg and 3 mL/kg in 20 mL portions in children weighing >20 kg.29 Negative suction pressure should be adjusted to avoid visible airway collapse. The first bronchoalveolar lavage fluid aliquot should be unfiltered and used for microbiological studies, and the other aliquots should be pooled (unless one is bloodstained), filtered through sterile gauze only if a lot of mucus is present, which is unlikely in chILD, and used for analysis of cellular and non-cellular components. Transport and preparation of samples is provided in our online SOP. Clinicians should liaise with their laboratories to ensure appropriate investigation of samples, to include appropriate microbiological analysis and culture (bacterial, fungal, virus), Oil Red O staining (fat-laden macrophages), periodic acid–Schiff staining (pulmonary alveolar proteinosis) and iron staining (ie, Prussian blue for haemosiderin within macrophages) to diagnose pulmonary haemorrhage syndrome. CD1a-positive cells are consistent with Langerhans cell histiocytosis. A lymphocyte-predominant BAL with an elevated CD4/CD8 ratio (>2) suggests sarcoidosis with involvement of the lung.30 Metabolic abnormalities may be diagnosed by BAL, for example, sea-blue macrophages.31

Endobronchial (EBx) and transbronchial (TBB) biopsies are rarely performed in ChILD and are not recommended unless a specific diagnosis that could be made by these techniques is suspected. Specific conditions such as pulmonary alveolar microlithiasis may be diagnosed by TBB,32 but the samples are too small to diagnose most chILDs. Blind EBx may detect sarcoid granulomas,33 but is not useful for diagnosing NEHI.34 Our protocols for these procedures can be found online if needed.

Invasive testing: surgical lung biopsy

The timing and need for lung biopsy is controversial. chILD patients who are well and thriving may not merit biopsy even if the CT scan appearances are not typical. Some would consider that an oxygen requirement warrants a diagnostic lung biopsy, while others would wait and see. There is clearly merit in performing an invasive procedure only if treatment will be changed as a result. Steroids are a mainstay of treatment for chILD and the timing of biopsy related to their initiation is often dictated by circumstance. Where possible, biopsy prior to steroid treatment is recommended (to minimise risk to wound healing and to expedite specific chILD treatments, ie, TNF-α antagonist infliximab combined with methotrexate for sarcoidosis, and cyclophosphamide for angiitis with granulomatosis). The dilemma posed by the sick patient with chILD who is on the verge of ventilation (and biopsy would most likely tip to requiring ventilation) or is unstable on a ventilator often dictates that a steroid trial before a biopsy may be appropriate. If the child is already ventilated, unless the ventilatory requirements are very high, a biopsy can safely be performed.

The site of biopsy should be guided by a recent CT chest. There should be liaison between the surgeon, pathologist and paediatrician. Any other procedures which may merit general anaesthesia (eg, bronchoalveolar lavage, mucosal biopsy, placement of a vascular access device or gastrostomy) should be carefully planned. The tip of the middle lobe and lingula should be avoided, and biopsy should preferably be from two sites, and sample areas of varying disease severity. Increasingly, biopsy is using video-assisted thoracoscopic surgery rather than a minithoracotomy. Whatever technique is used, the procedure must only be undertaken by an experienced surgeon, who is confident of obtaining adequate biopsies (at least 10×10×10 mm); a very superficial biopsy, which does not contain distal airways, may result in diagnostic error.

Samples should be rapidly transported to a waiting pathologist working in an accredited laboratory working to international standards.35 There should be access to light microscopy and common special stains, immunohistochemistry, genetics services, microbiology and virology services and photographic equipment. Table 2 describes the postbiopsy sample processing. The remaining tissue should be fixed in formalin. Table 3 describes the routine stains to be used. Details of transport of specimens can be found online.

Other sources of lung tissue

Postmortem and explanted lung tissue should be accessed if at all possible, and handled in the same way as a lung biopsy. Even if permission for autopsy is refused, it may be possible to obtain permission for percutaneous, transthoracic True-cut needle biopsies. DNA of the patient and both parents should be stored with the family's permission, if this has not already been done. Another valuable source of tissue is the explanted lungs after lung transplantation.

The role of peer review

chILD is a spectrum of very rare conditions, and no centre will see many cases. The diagnosis may be obvious, as, for example, a suspected surfactant protein gene mutation when known disease causing mutations are detected. However, in many cases the interpretation of CT scans and lung biopsy material is difficult. We believe that the best practice is to peer review these cases so that patients have the benefit of a consensus view from different specialists who can pool and increase their expertise. This is an integral part of our protocols. In chILD-EU, cases are peer reviewed at point of diagnosis and revisited at 12 monthly intervals to gauge the precision of diagnosis and patient outcome.

Proposed treatment protocols: the Delphi process

In the absence of any randomised controlled trial data, a Delphi process was undertaken to achieve a consensus on treatment protocols to harmonise treatment approaches. The full details are given in the online supplementary material. Briefly, we conducted a two-round web-based Delphi consensus of clinicians caring for chILD. Clinicians were identified from registered email addresses with national paediatric respiratory society groups, including the paediatric assembly of the American Thoracic Society. The questions were derived by the chILD-EU core management team, with adaptation by the lead representatives for clinicians in North America (Robin Deterding) and Australia (Adam Jaffe). The survey consisted of a scenario stem and related questions. All scenarios related to children with proven or highly suspected ILD. The survey was in English. Survey 1 consisted of 8 scenario stems with 75 responses required, survey 2 of 9 lead statements with 11 individual responses required. Clinicians were asked to score according to agreement: 1 Strongly Agree; 2 Agree; 3 Neutral; 4 Disagree; 5 Strongly Disagree; 6 Don't know.

Survey 1 was approached by 173 clinicians. Twenty-nine responders who provided fewer than 50% of responses were removed from the analysis. Data is presented from 144 clinician responders: Australia 14, Canada 3, France 5, Germany 15, Italy 1, USA 61, UK 37 and other 8. There were 127 respondents to the second round of Delphi. Twelve provided insufficient data responses (fewer than 50% responses) and so were removed from the complete analysis. The 115 consultant/attending pulmonologist respondents contributing to the second round of Delphi were from the UK (36), USA (30), Germany (22), Australia (9), France (6), Italy (5), Canada (2) and other countries (5).

Delphi 1 consisted of four scenarios related to the care of a patient who was very sick and ventilated or close to ventilation for whom no specific treatment was known. Delphi 2 provided statements derived from the consensus of opinion from the first Delphi round and asked how clinicians viewed the statement. Options, provided for varying doses and medicines, are summarised as (a) usual practice or (b) could happily use it if it is the consensus of the Delphi, or (c) not usual practice and would not wish to use it even if it is the consensus of the Delphi.

In all scenarios, we were clear that the suggested doses should be considered starting doses and could be changed at the discretion of the treating clinician at any time. The final consensus for treatment reached is summarised in table 4, and the consensus for a significant response in table 5.

Other potential therapies

There is even less evidence for other regimes and no recommendations are made. If there is evidence of systemic disease, and if cytotoxic or biological therapies are contemplated, it would usually be wise to involve a paediatric rheumatologist.

Monitoring chILD

The variety of ChILD diagnoses makes a single common monitoring plan of little value. chILD-EU has looked to enable reference across diagnoses by the development of an observational trial protocol focused on the first year of diagnosis. Monitoring is at months 1, 2, 3, 6 and 12 and annually thereafter. Key observations are clinical (respiratory rate, heart rate, weight, oxygen saturation in air awake, oxygen saturation including overnight while asleep and on exercise, evidence of pulmonary hypertension) and radiological monitoring (chest X-ray at diagnosis, 6 and 12 months; CT not recommended, however, if considered justifiable, a limited cut thin-section HRCT of areas of interest should provide sufficient information). In older children spirometry at each observational monitoring visit should be recorded, with DLCO (pulmonary haemorrhage monitoring) and body plethysmography recommended as indicated but at least once per year.

Pulmonary exacerbations

Acute pulmonary events associated with instability are a characteristic of many chronic diseases. Although recognised, they are often difficult to define. No definition exists for a pulmonary exacerbation in chILD and so we have sought, through the Delphi process and a review of the evidence, to define what is a clinically significant change in chILD as a first step towards a definition of a pulmonary exacerbation (table 5). The period of stability before and between exacerbations must be defined; for ChILD this is arbitrarily considered to be at least 7 days of stability at baseline values. More work is ongoing to define the definition of a pulmonary exacerbation and its relationship with other respiratory events such as upper respiratory tract infection and aspiration.

Summary and conclusions

We have proposed a sequence of investigations for chILD, and provided SOPs for performance of these tests. We aim to harmonise investigation of chILD to facilitate future research; this manuscript is the framework on which this will be based.