METHODS
Study populations
A total of 644 patients were treated with molecular targeted antineoplastic agents at Soonchunhyang University Seoul Hospital between May 2005 and September 2014. Targeted antineoplastic agents included mAbs (rituximab, cetuximab, bevacizumab, and trastuzumab) and TKIs (gefitinib, imatinib, erlotinib, and sorafenib). We searched for patients who underwent chest computed tomography (CT) or high-resolution CT (HRCT) scans during the course of targeted therapy. Of the 644 patients, 364 underwent chest CT or HRCT scans during the course of targeted therapy. Two radiologists (one chest radiologist with 15 years of experience in performing chest radiology and a junior resident in radiology) retrospectively reviewed the chest radiographs and CT scans of the 364 patients. After conducting a review of the radiological imaging, we identified 90 patients who presented with pulmonary complications on chest imaging. The medical records of these patients were reviewed by two clinical physicians in order to establish a diagnosis of drug-related pulmonary toxicity. The study was approved by the Institutional Review Board of Soonchunhyang University Hospital, Seoul, Korea (IRB No. 2013-01-008). Written informed consent by the patients was waived due to a retrospective nature of our study.
Patterns of radiologic pulmonary complications
Two CT scan devices (Sensation 16- and 64-section, Siemens Medical Solutions, Erlangen, Germany) were used in this study. The images were obtained volumetrically using multi-detector CT scan devices with 0.6-mm or 1-mm collimation and 1-mm (high-spatial-frequency algorithm) or 3-mm (standard algorithm) reconstruction.
The CT findings were interpreted based on the recommendations of the nomenclature committee of the Fleischner Society [
9,
10]. Chest CT or HRCT images were assessed for major radiographic abnormalities: (1) air-space abnormalities, including air-space consolidation and ground-glass opacity; (2) interstitial abnormalities, including intralobular and interlobular septal thickening, bronchovascular bundle thickening, and honeycombing; (3) nodules and micronodules; (4) airway abnormalities, including bronchial wall thickening, bronchial dilatation, and air trapping; (5) cardiovascular abnormalities, including cardiomegaly, pulmonary thromboembolism (PTE), and pulmonary arterial enlargement; and (6) other findings, including mediastinal lymphadenopathy and pleural effusion. Anatomical distribution was classified as central, peripheral, peribronchovascular, or random. The zonal predominance of abnormalities was assessed as upper, middle, lower, or random. The distribution of parenchymal opacification was divided into the following patterns: diffuse, focal, or patchy patterns.
The semi-quantitative scoring system was used to evaluate the different abnormalities, including air-space abnormalities, interstitial abnormalities, nodules, and micronodules. Each abnormality was scored according to the involved area of the lung parenchyma: grade 1, 1% to 25%; grade 2, 26% to 50%; grade 3, 51% to 75%; and grade 4, 76% to 100% of the total lung. Airway abnormalities were scored using the method mentioned above. Cardiomegaly was classified as follows: mild, cardiothoracic (C-T) ratio of 0.5 to 0.6, moderate, 0.6 to 0.7; and severe, > 0.7. Pulmonary arterial enlargement was defined as an increase in the diameter of the main pulmonary artery (> 30 mm) and right and left pulmonary arteries (> 15 mm). Lymphadenopathy was classified based on its short-axis diameter: mild, < 1 cm; moderate, 1 to 1.5 cm; and severe, > 1.6 cm. Pleural effusion was defined as the distance between the inner costal margin and meniscus of the pleural fluid on axial CT scan: small, < 1.5 cm; moderate, 1.5 to 4.5 cm; and large, > 4.5 cm. Evaluation of radiologic abnormalities and final diagnoses were made by the two observers.
Based on the CT analysis, targeted therapy-related pulmonary complications were categorized as follows: pneumonitis, nonspecific acute interstitial lung disease (ILD), organizing pneumonia, eosinophilic pneumonia, pulmonary fibrosis, bronchiolitis, pulmonary edema, adult respiratory distress syndrome (ARDS), diffuse alveolar hemorrhage (DAH), PTE, and pleural effusion. In cases of equivocal CT findings and lung parenchymal lesions with pleural or cardiovascular abnormalities, we selected one or more CT patterns. Each radiologic abnormality was defined as follows: (1) pneumonitis: interstitial abnormality with no remarkable lung fibrosis and less frequent air space abnormality, with CT patterns similar to that of idiopathic interstitial pneumonias, including non-specific interstitial pneumonia, desquamative interstitial pneumonia, and lymphoid interstitial pneumonia; (2) non-specific acute ILD: diffuse and less frequently patchy interstitial abnormality, which is mostly associated with ground-glass opacity and air-space consolidation and cannot be well categorized as the abovementioned pneumonitis pattern; (3) organizing pneumonia: similar CT features with cryptogenic organizing pneumonia of idiopathic form; (4) eosinophilic pneumonia: CT features reflecting Loeffler’s disease or chronic eosinophilic pneumonia; (5) pulmonary fibrosis: irregular reticular opacities with traction bronchial/bronchiectasis commonly accompanied by honeycombing and zonal predilection of peripheral, subpleural, and lower lung or diffuse involvement; (6) bronchiolitis: poorly defined centrilobular nodules or branching densities most easily identified in the lung periphery; (7) pulmonary edema: diffuse or patchy ground-glass opacities or consolidations and smooth interstitial thickening with dependent, perihilar, or lower lung predominance; (8) ARDS: diffuse or patchy ground-glass opacities or consolidations with geographic pattern and sometimes centrilobular or lobular opacities, and inconspicuous septal lines, which show dependent and basal predominance; and (9) DAH: patchy or diffuse ground-glass opacities or consolidations, frequently associated with poorly defined centrilobular nodules, and interlobular septal thickening developing over days after the acute episode. Preexisting lung diseases were also evaluated and included emphysema, nonspecific post-inflammatory sequelae, and known neoplasms.
Patient eligibility and case definitions
A diagnosis of drug-related pulmonary toxicity requires the presence of the following clinical conditions [
11,
12]: (1) new onset symptoms after administering targeted therapy; (2) lack of evidence of infectious cause; and (3) radiologic findings defined as pneumonitis or nonspecific acute interstitial disease by radiologists.
The exclusion criteria for this study were as follows: (1) evidence of infection based on the results of the investigation of the infectious cause and clinical symptoms or improvement after treatment with antibiotics; (2) history of ILD; (3) evidence of primary cancer progression; (4) history of discontinuation of previous targeted therapy due to the occurrence of another side effect; (5) suspected lung lesions detected on chest CT or HRCT before receiving targeted therapy; or (6) completion of the targeted therapy without clinical suspicion.
Data collection
The following demographic data, disease characteristics, and treatment details were reviewed: age, sex, Eastern Cooperative Oncology Group (ECOG) performance status at diagnosis, cancer diagnosis, type of targeted agent, treatment method, treatment duration, response, comorbidity, survival duration, and laboratory findings (complete blood count and differential counts, blood chemistry tests, and O2 saturation during blood gas analysis).
DISCUSSION
Cancer treatments have significantly changed with the improved understanding of cancer cell biology, which has led to the development of novel molecular targeted antineoplastic agents. As use of targeted agents becomes more frequent for cancer treatment, physicians must assess for possible toxicities. Since the first report of gefitinib-associated ILD was published in Japan [
13], pulmonary toxicities associated with the use of molecular targeted agents have rarely been reported. The exact incidence of targeted agent-induced pulmonary toxicities remains unclear, with many studies reporting a variable incidence of these adverse events, from 0.2% to 10.9% [
14]. In our study, the incidence of targeted agent-induced pulmonary toxicities was 2.3%, which is consistent with the findings from previous studies [
14,
15]. Our study included both mAbs and TKIs, and pulmonary complications occurred similarly in both groups. The drug-related pulmonary toxicity was found to be higher in TKIs than in mAbs, but no significant difference was observed.
Since a variety of pulmonary complications can be observed on chest CT images in patients treated with targeted therapy, drug-induced pulmonary toxicity should be diagnosed based on the clinical symptoms and imaging findings. Many patients develop symptoms, including dyspnea, cough, and fever, which are mostly confined to the lungs [
16]. The patterns and severity of clinical manifestations may differ depending on the patient’s underlying illness and the relevant drug [
15]. As previously reported in other studies, the most common symptom was dyspnea. This result suggests that physicians should be concerned about patients with newly developed respiratory symptoms during the course of molecular-targeted agent treatment.
A diagnosis of pulmonary toxicity should be considered in patients with a history of exposure to molecular targeted agents, who present with new or progressive respiratory symptoms [
17]. Most importantly, the diagnosis is made after excluding other possible causes, including infections, iatrogenic intra-alveolar hemorrhage, radiotherapy-induced adverse effects, transfusion reactions, and progression of the underlying neoplasm. This is because the prognostic and therapeutic implications of these factors are extremely different [
16,
18]. Physicians should perform immediate imaging studies (chest radiographs and CT scan), appropriate cultures, serologic tests, and bronchoscopy with bronchoalveolar lavage in patients who may be at risk [
14]. Although chest radiography is commonly performed in patients with respiratory manifestations as an initial diagnostic imaging modality, chest CT is more sensitive in detecting parenchymal lung diseases. Therefore, the recognition of various CT manifestations and patterns of targeted agent-induced pulmonary toxicities is essential for timely diagnosis and appropriate management of any adverse events.
Risk factors for drug-induced pulmonary toxicity have been identified in several retrospective studies. Many variable risk factors such as preexisting pulmonary fibrosis, poor performance status, previous thoracic irradiation, smoking, older age, male sex, or hypoalbuminemia have been suggested [
1,
11,
12,
19,
20]. In this study, older age was the only risk factor for drug-induced pulmonary toxicity. This is probably because the present study targeted a variety of drugs and diseases.
Pulmonary toxicity may respond to appropriate treatment; however, mortality has also been reported [
16]. Although no randomized controlled trials have evaluated the treatment benefit, immediate discontinuation of the targeted agent and administration of systemic glucocorticoids with supportive care resulted in an improvement in both symptoms and radiographic findings [
1,
14,
15]. The dose and duration of glucocorticoids varied in the literature. According to the severity of respiratory failure, high-dose methylprednisolone (500 mg to 1,000 mg/day for 3 days) was recommended [
7]. In this study, the administration of targeted agents was discontinued in all patients with or without treatment of systemic glucocorticoids; among them, 53.3% patients showed resolution of symptoms and improvement in radiological findings. Therefore, careful administration of systemic glucocorticoids, in accordance with the patient’s baseline comorbidities and other possible infectious conditions, is necessary. In addition, the beneficial impact of systemic glucocorticoids, according to relevant targeted agents, needs to be validated in future studies.
In our study, among the patients who discontinued taking the targeted drugs because of pulmonary toxicity, three were re-treated with equivalent drugs after resolution of clinical symptoms and radiological manifestations. Almost all patients who were treated with the same drug in combination with low-dose glucocorticoid did not experience a recurrence of pulmonary toxicity. However, a few studies have reported the re-administration of an equal molecular agent. Peerzada et al. [
14] reported that four (36.4%) among 11 patients had relapsed after re-treatment with imatinib, following the diagnosis of imatinib-induced pulmonary toxicity. Thus, physicians must carefully determine whether chemotherapy should be discontinued or not in patients receiving a molecular agent as the main therapy for cancer, and further studies are needed.
Previous clinical trial reports and the study summaries of these targeted agents have reported various life-threatening respiratory side effects. Approximately one-third of reported cases are fatal [
1]. Therefore, oncologists must assess for signs of toxicities early to prevent mortality. Targeted agent-induced pulmonary toxicity can be difficult to diagnose due to the variations in patients’ radiological findings. Therefore, the causative drug must be recognized, together with an understanding of the radiologic manifestations of pulmonary toxicity caused by these targeted agents, in order to provide the appropriate management [
17]. Oncologists, in conjunction with radiologists, should be knowledgeable of the clinical manifestations and variable radiologic manifestations of drug-induced pulmonary toxicity, which can increase in incidence with the treatment of diverse types of molecular targeted agents.
Drugs have been implicated in lung injury as a result of direct pharmacological action, persistence, or metabolism in the tissue, or via the production of a reactive metabolite. There is limited knowledge on how the timing of drug administration or drug interactions may interfere with the repair mechanisms or modulate the expression of pulmonary toxicity. Unlike conventional chemotherapy, where the actions are systemic and direct as results of biotransformation or cell injury, the use of targeted antineoplastic agents more likely affects the epithelia via a pharmacological effect. Some studies have reported the possibility that low albumin levels are a risk factor for mAbs or TKIs, especially gefitinib because a low albumin level is one of the markers of malnutrition [
11,
12]. Hence, future studies should identify other risk factors for the development of adverse pulmonary reactions and biological markers that can be used to predict the occurrence of pulmonary toxicity. Proteomics and genomics will hopefully identify the candidates for this role [
21].
Several limitations should be considered when interpreting our findings: the retrospective design of the study, the small number of patients, the heterogeneity of molecular targeted agents, and baseline diseases. In the future, a prospective study with a large number of patients treated with independent agents should be conducted.
In conclusion, the results from this study suggest that pulmonary toxicities associated with the use of molecular targeted agent therapy rarely occur but can be potentially fatal. Hence, physicians need to be aware of the varied clinical and radiological manifestations to provide appropriate management.