Abstract
Interstitial lung disease (ILD) secondary to drug-induced lung injury is an increasingly common cause of morbidity and mortality. The number of drugs associated with the development of ILD continues to rise, mainly due to the use of novel monoclonal antibodies and biologicals for neoplastic and rheumatological diseases, and includes, among others, chemotherapeutics, molecular targeting agents, immune checkpoint inhibitors, antibiotics, antiarrhythmics and conventional or biological disease-modifying antirheumatic drugs. Drug-induced ILD (DI-ILD) manifests with a variety of clinical patterns, ranging from mild respiratory symptoms to rapidly progressive respiratory failure and death. In most cases, there are no pathognomonic clinical, laboratory, radiological or pathological features and the diagnosis of DI-ILD is suspected in the presence of exposure to a drug known to cause lung toxicity and after exclusion of alternative causes of ILD. Early identification and permanent discontinuation of the culprit drug are the cornerstones of treatment with systemic glucocorticoids being used in patients with disabling or progressive disease. However, for certain drugs, such as checkpoint inhibitors, the frequency of lung toxicity is such that mitigation strategies are put in place to prevent this complication, and occurrence of DI-ILD is not necessarily synonymous with permanent drug discontinuation, particularly in the absence of valid therapeutic alternatives.
Abstract
Interstitial lung disease is a potentially severe and even fatal adverse drug reaction. The number of culprit drugs continues to increase. Identification and discontinuation of the causative drug is the cornerstone of treatment. https://bit.ly/3IApehy
Introduction
Drug-induced interstitial lung disease (DI-ILD) is a large and very heterogeneous group of adverse drug reactions, ranging from mild to progressive and life-threatening disease. More than 400 drugs have been reported to cause ILD [1], the most common being disease-modifying antirheumatic drugs (DMARDs), antiarrhythmics and antimicrobial and antineoplastic drugs [2], including immune checkpoint inhibitors (ICIs). The list of culprit drugs is constantly increasing, mainly due to the exponential development of antineoplastic drugs with specific mechanistic targets (including, among others, monoclonal antibodies and tyrosine kinase inhibitors), many of which are associated with lung toxicity.
Clinical, laboratory, radiological and histological features of DI-ILD are nonspecific and variable even with the same offending drug, suggesting that other factors, including genetics, may contribute to disease development. The diagnosis is one of exclusion and involves clinical suspicion, exposure to a drug known to cause lung toxicity and exclusion of other causes of ILD [3]. However, assessing causality is challenging, particularly when a potentially pneumotoxic drug (e.g. methotrexate) is used to treat a disease that can cause ILD (e.g. rheumatoid arthritis), in patients using multiple potentially pneumotoxic drugs (e.g. methotrexate and infliximab), or when drug toxicity occurs acutely in patients on chronic treatment with the culprit drug (table 1). Special attention should be paid to excluding other causes of ILD, including infection, heart failure or lymphangitic carcinomatosis (table 2). Naranjo et al. [4] developed an adverse drug reaction probability scale that can also be applied to suspected DI-ILD. They classified the probability that the adverse event is related to a given drug as “definite” (total score ≥9), “probable” (total score 5–8), “possible” (total score 1–4) or “doubtful” (total score ≤0) (table 3). Discontinuation of the culprit drug is the mainstay of treatment, whereas glucocorticoids are reserved for patients with disabling or progressive disease.
Main culprit drugs based on clinical and imaging patterns
Diagnostic checklist in patients with suspected drug-induced interstitial lung disease (DI-ILD)
Adverse drug reaction probability scale
Search strategy and selection criteria
We performed a comprehensive (but nonsystematic) literature search for articles related to DI-ILD. References included in this narrative review were identified by searching PubMed (https://pubmed.ncbi.nlm.nih.gov) for articles published up to December 2021, using the terms “interstitial lung disease”, “pulmonary fibrosis”, “lung injury”, “lung toxicity” OR “pneumonitis” AND “drug-induced”, “chemotherapy”, “antineoplastic agents”, “immune checkpoint inhibitors”, “mTOR inhibitors”, “antibiotics”, “nitrofurantoin”, “amiodarone”, “antiarrhythmic drugs”, “statins”, “disease-modifying antirheumatic drugs”, “methotrexate” OR “biological agents”. Relevant references cited in these articles were also screened. We limited our search to articles published in English and reviewed them manually. Articles in other languages with abstracts in English were also reviewed if sufficient detail was present in the abstract. The final reference list was generated based on the relevance to the topic covered in this review article.
This article outlines the main features of DI-ILD and provides a framework for approaching patients with this often under-recognised form of ILD.
Epidemiology
The incidence of DI-ILD is difficult to estimate, and varies widely depending on a number of factors, including the specific drug and dose, and the accuracy of reporting. In a retrospective cohort study of 770 consecutive Japanese patients diagnosed with advanced nonsmall cell lung cancer (NSCLC) between January 2004 and January 2014, 44 (6%) developed pneumonitis during systemic anticancer therapy with a mortality rate of pneumonitis of 36% [5]. More recently, Jo et al. [6] conducted a nested case–control study to identify associations between drugs with potential risk of ILD and the occurrence of DI-ILD in hospitalised patients requiring glucocorticoid therapy, using a national inpatient database (cases n=1541, controls n=5677). Of the 42 categories of drugs investigated, epidermal growth factor receptor (EGFR) inhibitors (OR 16.84) and class III antiarrhythmic drugs (OR 7.01) were those associated with the highest disease risk.
In cohorts of patients with ILD, between 3% and 5% of prevalent cases are drug-induced [7–9], corresponding to an incidence of DI-ILD ranging between 4.1 and 12.4 cases per million per year [10], but this is likely to be an underestimate, given the significant expansion of novel oncology drugs with a high rate of DI-ILD.
Clinical features
Disease onset varies from days to even years and is unpredictable. Acute pneumonitis manifests within hours or days with shortness of breath, fever and often peripheral eosinophilia, whereas symptoms and signs of pulmonary haemorrhage range from dyspnoea, cough and fever to acute respiratory failure, haemoptysis and acute anaemia, depending on disease severity [11]. In DI-ILD patients with subacute or chronic onset, or with subacute worsening of an underlying chronic respiratory disease, the main symptoms are worsening dyspnoea and reduced exercise capacity. Lung auscultation may reveal fine or “velcro-like” crackles, whereas digital clubbing is uncommon. In advanced disease, pulmonary hypertension and right ventricular dysfunction may occur [12]. Lung function test results may vary from an obstructive to a restrictive ventilatory defect with impairment of gas exchange depending on the underlying histopathological disease pattern.
Complementary diagnostic tests
Laboratory findings are nonspecific and may help to exclude alternative causes of ILD. However, white blood count may reveal increased eosinophils in DI-ILD manifesting as eosinophilic pneumonia and, rarely, hypersensitivity pneumonitis (HP) [2, 11]. Mild peripheral eosinophilia (0.5–1.5 g·L−1) may also accompany DI-ILD not presenting as eosinophilic pneumonia (i.e. amiodarone-induced pulmonary disease), thus raising the suspicion of DI-ILD. Bronchoscopy plays an important role in the diagnostic work-up of patients with suspected DI-ILD, mainly by ruling out alternative diagnoses, especially opportunistic infection and malignancy. The most prominent bronchoalveolar lavage (BAL) feature of DI-ILD is a lymphocytic alveolitis with a preponderance of CD8+ cells, but a preferential increase of CD4+ cells may be seen with the use of methotrexate, nitrofurantoin and sirolimus [13, 14]. BAL may also help define the underlying histopathological features of DI-ILD such as eosinophilic pneumonia (in the presence of BAL eosinophilia >25%), or alveolar haemorrhage.
Imaging
Thin-section computed tomography (CT) has a central role in the diagnosis of DI-ILD by defining the extent and distribution of the disease. In addition, HRCT may identify findings suggestive of alternative diagnoses. However, CT features are not specific for DI-ILD, and the same drug can be associated with multiple disease patterns, which can coexist in the same patient. A recent position paper from the Fleischner Society has proposed three key diagnostic criteria for DI-ILD, as follows. 1) New-onset lung parenchymal opacities at thin-section CT, commonly in a bilateral nonsegmental distribution; 2) temporal association of presentation with the administration of a systemic therapeutic agent; 3) exclusion of other causes of ILD [15]. This latter criterion may be particularly challenging, although common differential diagnoses of ILDs such as congestive heart failure, radiation pneumonitis or lymphangitic carcinomatosis are generally differentiated from DI-ILD based on CT features.
Overall, radiological patterns reflect the generally inflammatory/immunological nature of DI-ILD. Accordingly, the most common disease patterns include nonspecific interstitial pneumonia (NSIP), organising pneumonia, diffuse alveolar damage (DAD), HP and simple pulmonary eosinophilia [15]. Usual interstitial pattern (UIP), which is characterised on high-resolution computed tomography (HRCT) by reticular changes and honeycombing with or without traction bronchiectasis with a subpleural basal predominance, is a rare pattern of DI-ILD and its presence should raise the suspicion of alternative diagnoses, or pre-existing pulmonary fibrosis.
DAD, a common manifestation of DI-ILD, is characterised on HRCT by bilateral consolidation, ground-glass opacity (GGO) and crazy paving with posterior and basal predominance in the exudative phase [16]. In the organising and fibrotic phase of DAD, reticulation, traction bronchiectasis and architectural distortion may develop. Drug-induced HP manifests as diffuse GGO and ill-defined centrilobular nodules, and may be difficult to distinguish from HP induced by inhalation of organic dusts [17]. Areas of decreased attenuation on expiratory CT scans indicate air trapping, which is caused by bronchiolar obstruction [18]. HRCT abnormalities of drug-induced NSIP are also indistinguishable from those of idiopathic NSIP, and include GGO with varying degrees of reticular changes, consolidation and traction bronchiectasis, with subpleural sparing of the dorsal regions of the lung and with a lower lobe predominance [19]. Diffuse alveolar haemorrhage (DAH) may also manifest as extensive bilateral GGO. Organising pneumonia is another common radiological pattern of DI-ILD, manifesting mainly as multifocal areas of airspace consolidation that are peribronchovascular or peripheral in distribution and have a predilection for the lung bases [20].
There are no CT features that are specific for a drug aetiology or can differentiate DI-ILD from infection, including coronavirus disease 2019 pneumonia, or other non-DI-ILD. Moreover, radiographic abnormalities poorly correlate with the underlying histopathological pattern [17, 21]. Cleverley et al. [22] compared HRCT appearances, such as disease pattern and distribution, with histopathological features in 20 patients with biopsy-proven DI-ILD to determine the prognostic value of HRCT. The most common CT abnormalities were GGO (n=17), interlobular septal thickening (n=15), consolidation (n=14) and centrilobular nodules (n=8). The HRCT and histological patterns were concordant in only nine (45%) out of 20 cases. In addition, the HRCT pattern was of limited prognostic value. In many cases, the HRCT pattern is indeterminate, and it is likely that pathology, if available, would demonstrate cellular and possibly fibrosing ILD, often remaining difficult to classify. Radiological–histological correlation may be better in ILD induced by chemotherapy, particularly in bleomycin-induced lung injury, which on HRCT may manifest as NSIP, DAD, organising pneumonia, or fibrotic changes such as reticulation, traction bronchiectasis and honeycombing [21].
Histopathology
Drugs can produce a variety of histopathological patterns of interstitial pneumonia, ranging from acute/subacute to established fibrosis, and any histological pattern may be caused by a number of different drugs (table 4). Apart from a minority of drugs that may cause specific changes of lung morphology (e.g. direct cytotoxic damage with drug accumulation leading to foamy changes of intra-alveolar histiocytes and type II pneumocytes), the majority of DI-ILDs are generally reported by pathologists in descriptive terms that do not fit a unique histological pattern. The most common morphological patterns of DI-ILD include acute lung injury (DAD and organising pneumonia), cellular and/or fibrotic NSIP, HP, granulomatous pneumonitis, eosinophilic pneumonia, pulmonary haemorrhage or oedema, constrictive (obliterative) bronchiolitis and vascular modifications (e.g. veno-occlusive disease) [2]. Desquamative interstitial pneumonia (DIP), vasculitis and alveolar proteinosis are less common findings [23–25], the latter being observed mainly following tyrosine kinase inhibitor, sirolimus and everolimus treatment [26–28].
Histological patterns observed in drug-induced interstitial lung disease
DAD is characterised by an initial exudative phase with oedematous hyaline membranes and acute interstitial inflammation, followed by an organising phase with alveolar septal fibrosis and type II pneumocyte proliferation (figure 1a). Hyaline membranes consist of cellular debris, fibrin exudate and surfactant. DAD can be associated with organising pneumonia, progress to fibrosis or resolve with restoration of normal lung structure; therefore, different histological patterns might be appreciated depending on the timing of sampling (i.e. DAD in the early phase and organising pneumonia or NSIP in the late phases of DAD). Organising pneumonia is characterised histologically by excessive proliferation of granulation tissue, which consists of fibroblasts and myofibroblasts embedded in a myxoid-to-fibrotic stroma (so-called Masson bodies), involving alveolar ducts and alveoli (figure 1b). Cicatricial organising pneumonia with/without ossification is a newly described entity distinguished from conventional organising pneumonia by the presence of dense fibrous bands (figure 1c) [29]. Acute fibrinous and organising pneumonia and granulomatous and organising pneumonia are two additional variants of organising pneumonia that may be observed in drug-induced lung disease. Constrictive bronchiolitis with distortion of the bronchiolar lumen secondary to submucosal scarring and smooth-muscle hypertrophy may also be observed. NSIP, one of the most common histological patterns of DI-ILD, is characterised by relatively uniform chronic interstitial inflammation and type II pneumocyte hyperplasia in area of inflammation with or without interstitial fibrosis (figure 1d). Eosinophilic pneumonia is characterised by alveolar and interstitial eosinophilic infiltration and foci of organising pneumonia, but hyaline membranes and interstitial widening (i.e. DAD) can also be seen (figure 2a and b) [30]. Granulomatous pneumonitis may resemble other forms of granulomatous lung diseases such as HP, sarcoidosis, tuberculosis and nontuberculous mycobacterial disease (figure 3) [31–38]. Finally, a DIP-like pattern has been reported, mainly following sirolimus and nitrofurantoin therapy [39–43].
a) Diffuse alveolar damage. Thickened and oedematous interstitial space, hyperplastic pneumocytes and intra-alveolar fibrin deposition with scattered inflammatory cells in a patient with nonsmall cell lung cancer treated with pembrolizumab. b) Transbronchial lung biopsy showing organising pneumonia with intra-alveolar plugs consisting of active fibrosis (circled) on a background of nonspecific interstitial pneumonia in a patient with rheumatoid arthritis treated with tocilizumab. c) Cicatricial organising pneumonia characterised by polypoid lesions of dense collagenous fibrosis within the alveolar spaces with preserved lung architecture in a patient on chemotherapy (FOLFOX (folinic acid, fluorouracil, oxaliplatin) regimen) for colorectal cancer. d) Mixed fibrotic and cellular nonspecific interstitial pneumonia pattern with homogeneous interstitial fibrosis and lymphocytic infiltrate in a patient with advanced lung adenocarcinoma treated with nivolumab (transbronchial lung biopsy).
Eosinophilic pneumonia: a) marked eosinophilic infiltrate among macrophages (bronchoalveolar lavage), and b) alveolar fibrin exudate with organising pneumonia (transbronchial biopsy) in a patient treated with daptomycin.
Transbronchial lung biopsy showing a sarcoid-like reaction with marked lymphocytic infiltration and scattered non-necrotising granulomas (arrows) mimicking sarcoidosis in a patient treated with ipilimumab for metastatic melanoma.
The diagnosis of DI-ILD rarely requires a histological confirmation. In addition, morphological abnormalities of lung tissue (obtained generally by transbronchial biopsy) are not specific to drug-induced lung damage.
Pathogenesis and risk factors
The pathogenesis of DI-ILD is largely unknown for most drugs. However, the mechanisms through which drug-induced lung injury occurs are likely to involve direct damage to alveolar epithelial or capillary endothelial cells, dysregulation of the immune system, systemic cytokine release, cell-mediated lung damage and free-radical production with oxidative injury (table 5) [11].
Potential mechanisms of drug-induced lung damage
The development of DI-ILD is largely unpredictable, although cumulative dose and impaired renal function appear to confer an increased risk (table 6) [44]. Pre-existing ILD is a risk factor for the development of lung toxicity in patients with NSCLC treated with the EGFR tyrosine kinase inhibitors (TKI) gefitinib or erlotinib [45–47], and in patients with rheumatoid arthritis treated with leflunomide [48–51]. Conversely, whether pre-existing lung disease increases the risk of developing amiodarone lung toxicity is uncertain [52, 53]. Japanese ethnicity, male sex, smoking habit and poor performance status are additional risk factors for TKI-induced ILD in patients with lung cancer [26, 54, 55]. A seasonal distribution of ICI/TKI-induced pneumonitis has also been reported, suggesting that viral infection may act as a cofactor in the development of lung toxicity [56, 57].
Main risk factors for drug-induced interstitial lung disease (ILD)
Genetic factors have been suggested to contribute to the risk of developing DI-ILD. Udagawa et al. [58] carried out whole-genome sequencing of genomic DNA from 26 Japanese cancer patients who developed DI-ILD and identified associations with two intronic polymorphisms located within chromosome 22 open reading frame 34 (C22orf34; rs35198919) and teashirt zinc finger homeobox 2 (TSHZ2; rs12625311). Furthermore, in a subgroup analysis of patients with EGFR/TKI-induced ILD (n=13), an association between seven single nucleotide polymorphisms and DI-ILD was observed. Human leukocyte antigen (HLA) alleles have also been associated with increased risk of DI-ILD. Carriage of HLA-A*31:01 was significantly associated with methotrexate-induced ILD in Japanese patients with rheumatoid arthritis [59], while the combination of HLA-B*15:01 and DRB1*15:01 increased the risk of ILD in Japanese patients with advanced pancreatic cancer receiving gemcitabine plus erlotinib [60]. HLA-DRB1*04:05 has also been associated with DI-ILD, especially chemotherapeutic agents, in Japanese patients [61]. Notably, the higher allele frequency of HLA-DRB1*04:05 in the Japanese population compared to most other populations may account for the high susceptibility of Japanese patients to DI-ILD [62].
Treatment: general principles
Discontinuation of (and avoidance of further exposure to) the culprit drug is the mainstay of treatment. However, patients with disabling or progressive disease despite drug withdrawal are generally treated with glucocorticoids, although there are no robust data on the efficacy of glucocorticoid treatment in DI-ILD. Indeed, a systematic review that included 156 papers describing >6000 DI-ILD cases found that the majority of the data were of low or very low quality. The authors concluded that glucocorticoids are commonly used to treat DI-ILD and might be useful in severe cases [10]. Dose and duration of glucocorticoid therapy vary widely, mainly based on the radiological pattern of disease. Treatment response is highly variable, ranging from minimal or no improvement in patients with DAD to resolution in those with organising pneumonia. Recent data suggest that nintedanib (and possibly pirfenidone) is efficacious in patients with progressive fibrosing ILD despite appropriate treatment [63, 64]; however, whether these drugs are efficacious in patients with DI-ILD that progresses despite discontinuation of the culprit drug and glucocorticoid/immunosuppressive therapy is unknown.
In patients suspected to have idiopathic pulmonary fibrosis (IPF), a careful medication history, with emphasis on when the medication was started and stopped and its dosing, is a very important part of the diagnostic work-up. Based on how likely a given drug is to be the causative agent, its cessation may be indicated, although this does not always have an immediate effect on the patient's condition. Yet, disease progression following drug discontinuation is not expected in DI-ILD, whereas IPF is by definition a progressive disease. Moreover, in cases with a high pre-test probability of IPF (i.e. male patients, ex-current smokers, family history of IPF, age >60 years), the benefit of withdrawing a potentially pneumotoxic (but important) drug should be weighed against the risk of delaying a diagnosis of IPF. Lastly, and most importantly, UIP is an uncommon pathological and radiological pattern of DI-ILD. Therefore, the diagnosis of IPF and the initiation of antifibrotic therapy should not be delayed unnecessarily. With all these caveats, in patients suspected to have IPF, the decision to discontinue a medication should be taken on a case-by-case basis.
Prognosis
Owing to the large number of drugs that can potentially cause lung toxicity, prognosis is highly variable. Discontinuation of the culprit drug with or without glucocorticoids may lead to full recovery, provided the diagnosis is made early. Indeed, broadly speaking, early identification (and discontinuation) of the causative drug is generally associated with a favourable prognosis, whereas delayed diagnosis may lead to rapidly progressive acute respiratory distress syndrome (ARDS), or pulmonary fibrosis with poor prognosis. However, particularly with antineoplastic agents, the decision to discontinue the drug requires careful consideration of risks and benefits as well as the availability of alternative treatments. In selected cases of lung toxicity induced by checkpoint inhibitors, or mechanistic target of rapamycin (mTOR) inhibitors, continuation of the drug at a reduced dose may be justified in the absence of significant lung disease. Similarly, the decision on whether to rechallenge a patient with the same drug after a prior drug-induced lung toxicity must be made on a case-by-case basis with multidisciplinary discussion in a reference centre and based on the severity of the reaction and the availability of alternative therapies. DI-ILD recurs in approximately one-third of rechallenged cases [65, 66]. However, successful rechallenge (i.e. the safe and efficacious re-administration of a drug previously discontinued due to lung toxicity) after remission of severe DI-ILD has also been reported [67, 68].
Most common causative drugs
Chemotherapeutic agents
Lung toxicity has been reported to occur in 10–20% of all patients treated with antineoplastic drugs [2, 12, 69, 70]. Disease pathogenesis is poorly understood, but several mechanisms, either alone or in combination, are likely to be involved including direct damage to pneumocytes or alveolar endothelial cells, cell-mediated lung injury, oxidative stress, systemic cytokine release and dysregulated immune system in patients treated with ICIs [71].
Pulmonary toxicity induced by antineoplastic drugs typically occurs within weeks to a few months after treatment initiation and generally manifests as shortness of breath, cough and low-grade fever, although weight loss may also be present [71]. Chest auscultation may reveal bibasilar crackles. The most common pulmonary function abnormality is a reduced diffusing capacity of the lung for carbon monoxide (DLCO), with a restrictive ventilatory defect generally being observed in advanced or fibrotic disease [12]. Radiological abnormalities include patchy or diffuse GGO, consolidation, centrilobular nodules, interlobular septal thickening and reticular changes (figure 4a and b) [72]. Pleuroparenchymal fibroelastosis may be a late complication of treatment with alkylating agents (i.e. cyclophosphamide and carmustine) [73]. Bleomycin-induced lung damage may progress to end-stage disease with honeycombing [17] (figure 5), whereas hilar lymphadenopathy may be the presenting manifestation of methotrexate-induced lung toxicity [12]. Similar to other forms of drug-induced ILD, BAL reveals a lymphocytic alveolitis, although the main role of bronchoscopy is to exclude alternative diagnoses, mainly recurrent malignancy and infection. Lung biopsy has a limited role in the diagnosis of antineoplastic drug-induced lung toxicity, as there are no specific histological features and virtually all histopathological patterns of lung damage can be observed, including UIP, NSIP, organising pneumonia, DAD, alveolar haemorrhage, eosinophilic pneumonia and DIP [2]. As with other forms of drug-induced lung toxicity, drug withdrawal is the mainstay of treatment, although continuation of the drug at a reduced dose and drug rechallenge may be considered in selected cases. The use of systemic glucocorticoids is generally reserved to patients with severe/progressive pulmonary disease, but this treatment has not been evaluated in controlled clinical trials. The prognosis of chemotherapy-induced ILD is unpredictable, with severe or rapidly fatal outcomes despite drug discontinuation and glucocorticoid therapy being reported [74, 75].
Trastuzumab-induced interstitial lung disease in a patient with breast cancer. a) Computed tomography (CT) scan shows airspace consolidation in the right lower lobe. b) FOLFIRI (folinic acid, fluorouracil, irinotecan)-induced acute alveolar haemorrhage in a 52-year-old man with oesophageal cancer. Bilateral ground-glass opacities on high-resolution CT are due to subtotal alveolar filling with blood. Alveolar haemorrhage resolved rapidly following glucocorticoid treatment, but the patient died due to metastatic disease.
Bleomycin-induced interstitial lung disease. a) This 70-year-old woman with Hodgkin lymphoma was found to have new bilateral subpleural interstitial infiltrates following completion of a chemotherapy regimen containing bleomycin; b) over time (8 years), interstitial infiltrates progressed to end-stage disease despite glucocorticoid therapy.
Immune checkpoint inhibitors
The development of ICIs has revolutionised the treatment paradigm for cancer. ICIs have a broad range of indications including, among others, lung cancer, melanoma, bladder cancer and head and neck tumours, but they are also associated with high rates of pulmonary adverse effects [76]. Immune modulation resulting from checkpoint inhibition can lead to abnormal activation of autoreactive T-cells leading to inflammation in any organ system, although the precise mechanisms through which toxicity occurs remain to be established.
Interstitial pneumonias are described with every class of ICIs, with an incidence varying from 3% to 6%, including 1–2% of grade 3–4 adverse events [77–80]. The incidence of drug-induced ILD appears higher with programmed death-ligand 1 (PDL1) versus cytotoxic T-lymphocyte-associated protein 4 (CTLA4) inhibitors and increases in case of combined anti-PD1/PDL1 and anti-CTLA4 treatment (10%, all grades combined) [81]. Additional risk factors include a history of smoking, pre-existing ILD, chest radiation, poor performance status and treatment indication for lung cancer versus melanoma [81]. Interestingly, nivolumab may potentially be beneficial in patients with advanced NSCLC and pre-existing ILD, as recently reported in a real-world setting in France [82]. The onset of the iatrogenic lung toxicity is variable, from a few days to more than a year, with a median of 3 months [79, 83]. Clinical manifestations include cough, onset or progression of dyspnoea and chest pain, although ICI-induced lung toxicity may also be asymptomatic. Several patterns of radiological presentation have been reported, including organising pneumonia (the most frequent), NSIP, HP, DAD, alveolar haemorrhage and sarcoid-like reactions [84]. Importantly, tumour hyperprogression or pseudoprogression should be included in the differential diagnosis of ICI-associated ILD.
The management depends on the severity of lung disease [85], as follows. Grade 1 (mild: asymptomatic, radiographic abnormalities only): ICI withdrawal should be discussed, although the benefit/risk balance of maintaining immunotherapy will generally encourage continuing ICI. Grade 2 (moderate: symptoms without limitation of daily activities): requires ICI discontinuation; glucocorticoid treatment may be considered. Grade 3 (severe: symptoms with limitation of daily activities, or requirement of oxygen therapy): requires definitive drug discontinuation and glucocorticoid treatment. Grade 4 (life-threatening or disabling): requires definitive drug discontinuation and intravenous glucocorticoids; immunosuppressants may be needed. Grade 5 (fatal).
Rechallenge with the causative agent is possible in case of grades 1 and 2 toxicities, but not recommended for grades 3 and 4.
Adverse events of immunotherapy may also manifest as sarcoid-like pulmonary and cutaneous reaction and reactive hilar and mediastinal lymphadenopathy, which is often misdiagnosed as concomitant or progressive lung cancer [86, 87]. Biopsy of these lesions is generally indicated, and reveals typical nonnecrotising sarcoid-like granulomas. ICI-induced sarcoid-like reactions do not mandate treatment, especially if the condition is asymptomatic [88]. In cases requiring treatment, glucocorticoids with or without ICIs discontinuation appear to be effective [78–80, 88–90].
The management of patients with ILD and lung cancer should always be discussed in a multidisciplinary setting, weighing carefully the expected benefits and risks for each patient, particularly when ICIs are considered. However, we believe ICIs should be avoided in ILD patients with extensive disease and/or moderate/severe lung function impairment, particularly those with IPF, because of both the risk of acute exacerbations, and the lack of safety and efficacy data on co-administration of ICIs and antifibrotics (i.e. nintedanib and pirfenidone).
mTOR inhibitors
mTOR inhibitors exert immunosuppressive properties by reducing T- and B-cell proliferation [91], and are widely prescribed to prevent solid organ transplant rejection and for treatment of cancers and lymphangioleiomyomatosis. mTOR inhibitors are significant inducers of lung toxicity, which manifests mainly as lymphocytic interstitial pneumonia, organising pneumonia or alveolar haemorrhage [92]. BAL is typically lymphocytic, sometimes with increased eosinophil count, or haemorrhagic. Outcome is usually rapidly favourable following drug withdrawal or dosage reduction. Glucocorticoids may be necessary. Notably, sirolimus-induced pneumonitis may improve after switching to another mTOR inhibitor, everolimus [93].
Antibiotics
Nitrofurantoin (a 5-nitrofuran derivative) is commonly used for the treatment and prophylaxis of urinary tract infections (UTIs). The acute form of pulmonary toxicity accounts for ∼80% of cases and develops following a short course of therapy, although the disease may also develop insidiously following months or even years of treatment [94]. Nitrofurantoin-induced pulmonary toxicity occurs almost exclusively in women (more commonly middle-aged or elderly) because of their increased susceptibility to recurrent UTIs and more frequent use of the drug. The acute form of pulmonary toxicity results from a hypersensitivity reaction (type I or III), whereas cell-mediated or toxic responses have been proposed as the main pathogenetic mechanisms in chronic disease. Additionally, histopathological findings are different; indeed, acute disease is characterised by mild (and often eosinophilic) interstitial inflammation, whereas diffuse interstitial pneumonia with an NSIP pattern is commonly observed in chronic reactions [94–96].
Acute pneumonitis generally develops within 1–2 weeks of nitrofurantoin use (or earlier in case of previous exposures) and presents with fever, dyspnoea and cough; peripheral eosinophilia is also common [94, 97]. In subacute and chronic disease, the most common presenting symptoms are dyspnoea and cough, which develop after ≥1 month of treatment [96]. Radiologically, acute pulmonary toxicity manifests as diffuse GGO, whereas GGO and consolidation with or without reticular changes and traction bronchiectasis is the most common pattern in patients with chronic disease [98]. Pleural effusion may be seen in acute disease, but is an uncommon finding in chronic reactions. Autoimmunity can be present in the form of antinuclear antibodies (ANA) or antineutrophil cytoplasmic antibodies.
Prompt discontinuation of nitrofurantoin is the cornerstone of therapy for both acute and chronic lung disease [99], but while in acute pneumonitis symptoms and radiographic abnormalities improve rapidly, over days to weeks [94, 97, 100], chronic toxicity may require weeks to months to resolve. The use of glucocorticoids is generally limited to patients with severe and progressive disease, although resolution of severe pulmonary toxicity may occur without treatment [101, 102]. Overall, the prognosis is favourable, but fatal ILD has also been reported [94, 97, 103]. Drug rechallenge should be discouraged, as it invariably causes disease relapse.
Drug rash with eosinophilia and systemic symptoms
Drug rash with eosinophilia and systemic symptoms (DRESS) syndrome is a severe idiosyncratic drug reaction characterised by extensive skin rash with frequent facial oedema, polyadenopathy, blood eosinophilia and a wide range of organ involvement, including hepatitis and nephritis [104]. Lung involvement (i.e. pulmonary infiltrates and pleural effusion) is present in ∼20% of cases and is a major cause of morbidity. Fatalities caused by severe lung disease have also been reported [105]. More than 80 medications have been associated with DRESS syndrome, the most frequently incriminated being antibiotics, but also anticonvulsants, allopurinol, nonsteroidal anti-inflammatory drugs and antidepressants. The latency from drug initiation to symptom onset typically ranges from 2 to 8 weeks. The pathogenesis of DRESS syndrome is poorly understood, but host factors (i.e. carriage of certain HLA alleles) and viral infection, mainly human herpesvirus-6, are believed to be involved [106].
Cardiovascular drugs
Amiodarone lung
Amiodarone is an iodine-containing compound commonly used to treat supraventricular and ventricular arrhythmias. Up to 5% of patients may develop pulmonary toxicity [12, 107–109], which may manifest as interstitial pneumonia, organising pneumonia, ARDS, eosinophilic pneumonia and, rarely, diffuse alveolar haemorrhage, lung nodules and masses [110].
Interstitial pneumonia
Interstitial pneumonia, the most common manifestation of amiodarone-induced lung toxicity, generally develops within 6–12 months of treatment, although cases of lung disease occurring within few weeks or after several years of treatment have also been reported [12, 111, 112]. Risk factors for pulmonary toxicity include a daily dose >400 mg, long-term treatment and age ≥60 years [113]; conversely, whether pre-existing lung disease increases the risk of developing lung toxicity remains controversial. Disease pathogenesis is incompletely understood, although both a direct cytotoxic effect and an indirect immunological reaction are believed to be involved [114, 115]. An additional pathogenetic hypothesis postulates that the drug induces alveolar cell apoptosis by acting as a nonselective thyroid-hormone receptor antagonist, thus disrupting the thyroid hormone signalling pathway. In addition, amiodarone accumulates in adipose tissue, thus exhibiting an increased half-life [108, 116]. The mode of presentation is insidious in ∼55% of cases (over a period of 1–3 months) and rapidly progressive in 40% of cases (acute amiodarone pneumonitis); the disease may be asymptomatic in 5% of cases. Dyspnoea and nonproductive cough are the most common presenting symptoms [12, 114, 115]. Fever is present in up to 50% of cases, whereas weight loss, malaise and pleuritic chest pain are less common manifestations [114, 115]. Inspiratory crackles may be heard on chest auscultation. Laboratory abnormalities are nonspecific and blood levels of amiodarone are generally normal [114, 115].
HRCT reveals areas of increased attenuation in the lungs (wherein infiltrates may be diffuse, unilateral or predominant on one side), but also in the liver and spleen, owing to the tendency of amiodarone to accumulate in tissue macrophages (figure 6a and b) [117]. Additional HRCT features include diffuse GGO and thickened interlobular septa; traction bronchiectasis and honeycombing can also be observed (figure 6c and d) [118, 119]. A restrictive ventilatory defect with reduced DLCO is the most common functional abnormality. Bronchoscopy with BAL is more helpful in ruling out alternative diagnoses, such as infection or malignancy, than confirming the diagnosis of amiodarone lung toxicity, as the BAL cellular pattern is nonspecific. The presence of “foamy” macrophages, which is due to the accumulation of phospholipids in alveolar macrophages, although typical, is not pathognomonic of pulmonary toxicity, as these cells can be found in up to one-half of patients receiving amiodarone [114, 115]. Conversely, in the absence of foamy macrophages, the diagnosis of amiodarone lung is unlikely [114, 115].
Amiodarone-induced interstitial lung disease. a, b) 48-year-old man with atrial fibrillation. Lung window images from axial computed tomography (CT) show a) extensive ground-glass opacities identified after 2 years of treatment with amiodarone; b) after 1 year, despite drug discontinuation and initiation of glucocorticoids, ground-glass opacities became more extensive. c, d) 77-year-old man with atrial fibrillation. High-resolution CT c) before and d) 1 month after initiation of amiodarone. Axial CT shows extensive ground-glass opacities and increased interstitial markings.
A clinical diagnosis of amiodarone-induced lung toxicity is supported by the following features: insidious onset of dyspnoea and/or cough; new GGO or reticular abnormalities on chest radiography/HRCT; presence of foamy macrophages in the BAL; exclusion of other causes of lung disease; and clinical and radiological improvement following amiodarone discontinuation (with or without glucocorticoids). Treatment consists primarily of amiodarone discontinuation and, in symptomatic patients, systemic glucocorticoids. However, because of the long half-life of amiodarone and its accumulation in fatty tissues, pulmonary disease may progress despite drug withdrawal. Overall, improvement is slow and disease recurrence is possible even weeks or even months after the drug has been discontinued [120]; therefore, low-dose glucocorticoids are often maintained for a few months even after improvement or recovery has been obtained. In patients who have experienced amiodarone lung toxicity, the drug should not be reintroduced because of the high risk of disease recurrence potentially more severe than the first episode.
Eosinophilic pneumonia
Amiodarone lung toxicity may manifest as acute (more commonly) or chronic eosinophilic pneumonia. Acute eosinophilic pneumonia (AEP) manifests acutely with fever, dyspnoea and dry cough, whereas in chronic eosinophilic pneumonia (CEP) respiratory symptoms may be accompanied by weight loss and night sweats [121]. Peripheral blood eosinophilia may be observed in both AEP and CEP. On chest HRCT, diffuse GGO and consolidation, which, in CEP, tend to distribute peripherally (“photographic negative” of pulmonary oedema), are the most common radiological abnormalities [122]. The diagnosis is supported by the presence in the BAL of foamy macrophages with eosinophilia >25%, and requires the exclusion of other causes of eosinophilic pneumonia (e.g. parasitic and fungal infection) and vasculitis. Treatment consists of drug discontinuation, with systemic glucocorticoids being reserved to patients with symptomatic or progressive disease.
Organising pneumonia
Organising pneumonia occurs in about one-quarter of cases of amiodarone lung toxicity and manifests as dyspnoea, dry cough and fever; pleuritic chest pain may also be present. Chest radiography typically reveals patchy consolidation, often with an air bronchogram, which mimics bacterial pneumonia, whereas HRCT shows in addition GGO and septal thickening [123]. Opacities tend to migrate. A histological confirmation of the diagnosis may be needed to when organising pneumonia manifests as chronic consolidation or mass. Histologically, organising pneumonia is characterised by excessive proliferation of granulation tissue (loose collagen-embedded fibroblasts and myofibroblasts) involving alveoli and alveolar ducts. Treatment consists of drug discontinuation generally associated with systemic glucocorticoids.
Acute respiratory distress syndrome
ARDS is a rare but severe form of amiodarone lung toxicity that has been reported in patients undergoing surgery or pulmonary angiography [124–127]. ARDS is defined by the acute onset (≤1 week) of respiratory failure, bilateral infiltrates on chest radiograph, hypoxaemia as defined by a arterial oxygen tension/inspiratory oxygen fraction ratio ≤200 mmHg and no evidence of cardiogenic oedema or fluid overload. More common causes of ARDS, such as sepsis, aspiration, transfusion and drug toxicity need to be excluded. In patients with ARDS, the main role of BAL is to rule out infection, haemorrhage and malignancy. The histological pattern of ARDS is DAD. Disease management includes drug discontinuation, supportive measures and mechanical ventilation. Most patients are also treated with systemic glucocorticoids. ARDS secondary to amiodarone toxicity has a mortality rate of ∼50% [128].
Diffuse alveolar haemorrhage
Amiodarone lung toxicity may rarely manifest as DAH, which tends to occur after few days or months of treatment, generally in patients with pre-existing pulmonary, cardiac or renal disease [129–131]. DAH presents acutely with cough, dyspnoea and fever; haemoptysis may also be present. Chest radiograph reveals diffuse and bilateral GGO and consolidation, whereas BAL is characterised by progressively more haemorrhagic returns [132]. Increased levels of hemosiderin-laden macrophages in BAL is an additional typical finding. The diagnosis of amiodarone-induced DAH requires the exclusion of other causes of alveolar bleeding, such as vasculitis, systemic lupus erythematosus (SLE), drugs other than amiodarone, or inhaled toxins. Similar to other forms of amiodarone lung toxicity, treatment consists of drug withdrawal and systemic glucocorticoids are the mainstay of treatment. Additional drugs that can induce DAH include, among others, anticoagulants, antiplatelet agents, new direct oral anticoagulants and thrombolytic agents.
Statins
Statins are lipid-lowering medications that are widely used for the prevention and treatment of cardiovascular disease. While generally safe and well tolerated, they can cause muscle aches or cramps in ∼5% of patients. Whether statin use increases the risk of developing ILD remains controversial. Indeed, while a systematic review of the literature and the United States Food and Drug Administration adverse event reporting database suggested that the use of statins may be associated with an increased risk of ILD [133], a large cohort study that included users of respiratory medications (>1.4 million patients, of which 6665 possible or probable cases of ILD) did not find an association between statin use and incidence of ILD [134].
Disease-modifying antirheumatic drugs
DMARDs are medications with immunosuppressive and immunomodulatory properties that are used for the treatment of a number of conditions including rheumatoid arthritis, psoriatic arthritis, ankylosing spondylitis and SLE. DMARDs are commonly classified as conventional or biological. The conventional DMARDs most commonly used are methotrexate, hydroxychloroquine, leflunomide and sulfasalazine, whereas biological DMARDs include, among others, infliximab, adalimumab, etanercept, rituximab, abatacept, tocilizumab and tofacitinib. Pulmonary toxicity has been reported with virtually all of the DMARDs, but the frequency and pattern of lung disease vary widely, based on the drug involved [1].
Methotrexate is the DMARD most commonly used in rheumatoid arthritis, but it is also used the treatment of other rheumatological diseases, psoriasis and some malignancies. Lung toxicity tends to occur after weeks to months of low-dose therapy, but has also been reported after short treatment with higher doses [135, 136]. The frequency with which methotrexate-induced lung disease occurs is difficult to estimate due to its nonspecific clinical, radiological and histological features, the concomitant use in the same patient of multiple potentially pneumotoxic drugs, and the possible presence of coexisting or pre-existing lung involvement from the disease for which methotrexate is used. However, in a systematic review of the long-term safety of methotrexate monotherapy in 3463 rheumatoid arthritis patients receiving the drug up to 36.5 months, only 15 (0.43%) cases of methotrexate-induced pneumonitis occurred [137]. Conversely, in patients with rheumatoid arthritis related ILD (RA-ILD), methotrexate does not seem to exert deleterious effects, and possibly is beneficial [138]. In a retrospective, case–control study of rheumatoid arthritis patients with (n=410) or without (n=673) ILD, methotrexate use was associated with a decreased risk of ILD [139], although, due to the retrospective design of the study, it cannot be excluded that some confounding factors might have affected this finding. Moreover, in a previous study of RA-ILD patients, methotrexate treatment was strongly associated with longer survival after adjusting for confounding variables [140], suggesting that the increased risk of ILD in rheumatoid arthritis patients treated with methotrexate reported in older studies might be accounted for, at least in part, by cases of pre-existing RA-ILD misdiagnosed as methotrexate-induced lung fibrosis [141]. Overall, there are no high-quality data suggesting that methotrexate is associated with chronic pulmonary fibrosis. HP is the most common form of methotrexate-induced lung toxicity and is characterised by lymphocytic (and, less commonly, eosinophilic) interstitial infiltration and poorly formed granulomas [142]. However, organising pneumonia, acute interstitial pneumonia and pleural effusion have also been described following methotrexate use [66]. As with most drugs, the mechanisms through which methotrexate-induced lung injury occurs are largely unknown, although a direct toxic effect of the drug or a hypersensitivity reaction may be involved [143]. Factors that increase the risk of methotrexate-induced lung toxicity include older age (e.g. >60 years), pre-existing lung disease, previous use of DMARDs and diabetes [144]. Methotrexate-induced lung toxicity may present in an acute, subacute (the most common) or chronic form. Acute pneumonitis manifests with rapidly progressive (over several days) dry cough, dyspnoea, fever, malaise and chest pain, whereas subacute pneumonitis is characterised by a more insidious onset. Mild peripheral eosinophilia is present in up to 50% of patients.
Chest radiograph shows diffuse (nodular or ill-defined) parenchymal infiltrates and, in severe cases, bilateral consolidation with air bronchogram, while HRCT displays patchy, widespread or diffuse GGO with or without consolidation or septal lines, and poorly defined centrilobular nodules [145, 146]. Hilar or mediastinal lymphadenopathy and pleural effusion are less common findings. Pulmonary function tests typically show a restrictive ventilatory defect with impaired gas exchange. As with other drug-induced lung diseases, bronchoscopy with BAL is more helpful in excluding alternative diagnoses, such as infection, than in diagnosing methotrexate-induced pulmonary toxicity. However, BAL shows increased cellularity and is typically lymphocytic with an increased CD4/CD8 ratio [147, 148]. Cellular interstitial infiltration (with or without granulomas) and acute and organising DAD are the main histological abnormalities, but lung biopsy is rarely required to confirm the diagnosis [66]. The diagnosis of methotrexate-induced pulmonary toxicity is based on a combination of clinical, radiological, BAL and histological (when available) features, exclusion of alternative causes of lung disease and clinical response to drug discontinuation.
Cessation of methotrexate may be sufficient for clinical improvement and even disease reversal to occur, whereas patients with severe or progressive disease despite drug withdrawal generally require glucocorticoid treatment. Overall, the prognosis of methotrexate pneumonitis is favourable; however, persistent radiological and functional abnormalities and even fatalities have been reported [66, 149, 150].
Biological agents
Biological agents are an increasingly recognised cause of DI-ILD, and tumour necrosis factor (TNF) inhibitors represent the most common causative drugs (apart from the increased incidence of pneumonia including tuberculosis reactivation) (table 7). ILD induced by biological therapies is relatively rare, although the exact prevalence of the disease is unknown [151]. Similar to other forms of DI-ILD, the most common presenting manifestations are nonproductive cough, dyspnoea and pulmonary infiltrates on chest radiograph. Perez-Alvarez et al. [152] reported on 122 cases of new-onset or exacerbation of ILD following biological therapies. The drugs associated with ILD were almost exclusively anti-TNF agents (e.g. etanercept in 58 cases and infliximab in 56 cases) and were used for treatment of rheumatoid arthritis in most cases (108 out of 122; 89%) [152]. DI-ILD was confirmed histologically in 26 cases: UIP (seven cases), NSIP (six cases) and organising pneumonia (five cases) were the most common pathological patterns. Notably, drug discontinuation (with or without glucocorticoids) led to ILD resolution or improvement in only two-thirds of patients, with mortality being particularly high among patients with pre-existing ILD [152]. Older age (e.g. >65 years) and concomitant immunosuppressive drugs were additional factors associated with increased mortality. Therefore, when considering a biological treatment, the benefit of the drug should always be weighed against the potential risk of pulmonary toxicity, particularly in elderly patients with rheumatoid arthritis and with a UIP pattern of disease. Sarcoid-like granulomatosis, pulmonary haemorrhage and organising pneumonia have also been described with TNF inhibitors. Sarcoid-like granulomatosis affects more frequently the lung (followed by the skin), and etanercept is the drug most often incriminated [153]. Following drug discontinuation, symptoms (cough, fatigue and dyspnoea) and radiographic abnormalities (especially mediastinal lymphadenopathy) tend to improve spontaneously over 2–6 months. In addition, TNF inhibitors can induce ANAs [154]. The majority of patients with positive ANAs are asymptomatic, but vasculitis and lupus-like syndromes have been described [155]. Pleuropulmonary involvement may occur, but less frequently than in other forms of drug-induced lupus.
Main respiratory side-effects induced by biotherapies and other promising inhibitors indicated for the treatment of autoimmune diseases
The anti-CD20 rituximab is widely used for treatment of malignant lymphoma and various autoimmune disorders including rheumatoid arthritis. Acute lung injury with bilateral infiltrates has been described within the first 24 h after the first injection, mostly in patients with neoplastic/haematological disorders, and may be fatal [156]. The mechanism of lung damage is unknown, but could be linked to a massive release of cytokines from lysed neoplastic cells. Most cases of rituximab-induced ILD occur on average 3 months after the first rituximab infusion and 2 weeks after the last rituximab infusion at the time glucocorticoids are withdrawn [156, 157]. Bilateral alveolar infiltrates and organising pneumonia may be observed on chest imaging. Withdrawal of rituximab along with glucocorticoids is usually very effective. Glucocorticoids given prophylactically to prevent the production of anti-rituximab antibodies may also reduce the rate and severity of drug-induced adverse events, including acute infusion reactions, cytokine release syndrome and lung toxicity [158]. Rechallenge should be avoided.
Conclusion
DI-ILD is a wide and highly heterogeneous group of conditions, and the list of culprit drugs is constantly increasing. The clinical, radiological and histopathological features of DI-ILD, though suggestive, are not specific; therefore, the diagnosis requires a high index of suspicion and the exclusion of alternative causes of ILD. Knowledge of the most common causative drugs and a diagnostic framework are also crucial. However, the diagnosis can be particularly challenging when multiple potentially pneumotoxic drugs are used for the treatment of diseases that frequently cause pulmonary complications, such as connective tissue diseases. Identification and discontinuation of the culprit drug (with or without glucocorticoids) is the cornerstone of treatment (figure 7), although for some agents, such as checkpoint inhibitors, the potential benefit of the drug and the frequency of lung toxicity are such that the occurrence of DI-ILD is not necessarily synonymous with drug discontinuation, particularly in the absence of valid therapeutic alternatives.
Management of patients with drug-induced interstitial lung disease. #: careful continuation of antineoplastic drugs/immunotherapy may be considered (grade 1 and grade 2 toxicities), particularly in the absence of valid alternative treatment options; ¶: multidisciplinary discussion in a reference centre highly recommended; +: depending on disease severity.
Shareable PDF
Supplementary Material
This one-page PDF can be shared freely online.
Shareable PDF ERJ-02776-2021.Shareable
Acknowledgements
P. Bonniaud would particularly like to warmly thank Philippe Camus (Dept of Respiratory Medicine and Critical Care, Hôpital du Bocage, Dijon, France) for his teaching and incomparable experience in the field of drug-induced lung diseases.
Footnotes
Conflict of interest: P. Bonniaud reports research grants (paid to his institution) from AstraZeneca, personal fees for participation to advisory board meetings from Roche, Boehringer Ingelheim, AstraZeneca and Novartis, and support for attending medical/research meetings from Roche, Boehringer Ingelheim, AstraZeneca, Novartis, Chiesi, Sanofi and Stallergenes. P. Spagnolo, G. Rossi, N. Sverzellati and V. Cottin have nothing to disclose.
- Received October 22, 2021.
- Accepted March 14, 2022.
- Copyright ©The authors 2022. For reproduction rights and permissions contact permissions{at}ersnet.org
References
