The incidence of pneumonitis in patients treated with ICIs varies in different analyses. One large meta-analysis involving 16 trials and 6,208 patients reported an incidence of 4% for any pulmonary toxicity and 1.5% for high-grade pneumonitis, which appeared similar across PD-1 and PD-L1 inhibitors (Sun et al., 2019). A subsequent, larger meta-analysis showed that anti-PD-1/PD-L1 was associated with a 5-6 fold higher risk of pneumonitis compared with cytotoxic chemotherapy (Huang et al., 2019). Notably, the odds ratio for pneumonitis for the combination of ipilimumab/nivolumab was approximately twice as high as single-agent anti-PD-1 therapy. Real-world studies have suggested similar incidences in NSCLC and melanoma patients, at approximately 5% for anti-PD-1/PD-L1 monotherapy and 10% for combination ipilimumab and nivolumab (Naidoo et al., 2017). Although fatal toxicities are uncommon (approximately 0.4% for single-agent anti-PD-1/PD-L1 and 1.2% for combination ipilimumab and nivolumab), pneumonitis likely comprises the most frequent fatal event, accounting for 35% of fatal toxicities from anti-PD-1/PD-L1 (Wang et al., 2018).

Interestingly, combinations involving anti-PD-1/PD-L1 with either chemotherapy or vascular endothelial growth factor (VEGF) inhibitors may actually have a slightly lower incidence of pneumonitis, although these regimens have generally not been directly compared with anti-PD-1 monotherapy. The meta-analysis described above suggested that the combination of pembrolizumab and chemotherapy had a lower risk of pneumonitis compared with anti-PD-1 therapy alone (although higher than chemotherapy) (Huang et al., 2019). Randomized studies, however, have shown that pneumonitis occurs in 5–8% of patients treated with pembrolizumab and chemotherapy, which appears broadly similar to that observed with single-agent therapy (Gadgeel et al., 2020; Paz-Ares et al., 2020). Pneumonitis appears perhaps slightly less common for the combination of pembrolizumab and axitinib (3%, with one fatal case) compared with single-agent anti-PD-1/PD-L1, although these have not been directly compared prospectively (Atkins et al., 2021).

To gain insight into the mechanisms of alveolar damage in ICI-pneumonitis, we will briefly discuss the mechanisms of non-ICI associated types of pneumonitis. Pneumonitis is a condition that encompasses a diverse spectrum of acute and chronic inflammatory etiologies that evoke immune-mediated damage to alveoli with resultant respiratory dysfunction. Triggers of pneumonitis include a variety of airborne organic and inorganic irritants, as well as radiation therapy, aspiration and medications delivered by any route, resulting in non-IgE-mediated hypersensitivity pneumonitis (HP, also known as extrinsic allergic alveolitis), radiation-induced lung injury (RILI), and drug-induced interstitial lung disease (DILD), respectively. Pneumonitis may vary with respect to severity and duration (sub-acute, acute, and chronic) and the more severe forms of these pulmonary disorders may lead to development of extensive fibrosis as a complication of disease progression. Acute interstitial pneumonitis (AIP, Hamman-Rich syndrome), on the other hand, is a distinct entity of unknown etiology that is rapidly progressive. It is an uncommon condition that clinically resembles the acute respiratory distress syndrome (ARDS) and is associated with a high mortality of up to 60% (Mastan et al., 2018).

Although not entirely clear, the immunopathogenesis of HP appears to involve Th17 cells that are activated following exposure to the offending irritant/antigen in the lungs. The resultant production of IL-17A promotes IL-8-mediated influx of neutrophils into the lungs with resultant inflammation-mediated alveolar damage (Pardo et al., 2000; Girard et al., 2011; Hasan et al., 2013; Costabel et al., 2020; Elmore et al., 2020). Neutrophil recruitment to the lungs may be exacerbated by low numbers of pulmonary regulatory T cells (Tregs) (Girard et al., 2011; Elmore et al., 2020) and by the production of IgG antibodies to the offending stimulus (Costabel et al., 2020).

Some patients with HP develop chronic, severe disease that is associated with the occurrence of fibrosis. Although the underlying immunopathogenesis remains to be established, an association with co-existent autoimmune disease has been described. In this context, one study has reported a high prevalence of autoimmune thyroiditis in patients with chronic HP (25.6% in patients vs. 10.7% in matched control subjects, p < 0.0001), which was found to be a significant predictor of mortality (HR = 3.39, p = 0,012) (Adegunsoye et al., 2017). Similar findings have been reported in a subset of HP patients who expressed the DRB1*03:01-DQB1*02:01 HLA haplotype that was associated with a significant risk of development of antinuclear antibodies (OR = 19.23; p = 0.0088), while expression of the HLA-DRB1*03:01 allele was associated with higher mortality (OR = 5.9; p = 0.043) (Buendía-Roldán et al., 2020). Two other recent studies have described associations between seropositivity for antinuclear antibodies in patients with HP that was correlated with disease severity and progression (Adegunsoye et al., 2015; Bonella et al., 2019).

Giuranno et al. (2019) have recently described in detail events involved in the pathogenesis of RILI. Among others, the most prominent mechanisms of pulmonary damage involve radiation-associated generation of reactive oxygen species (ROS), which are cytotoxic for both type I and type II pneumocytes (Giuranno et al., 2019). Type I cells line the alveoli and maintain structural integrity, while type II pneumocytes produce the protective, lower airway lining lipoproteins, surfactants A and D. In addition, radiation-induced damage to resident cells in the airways also results in the release of damage-associated molecular patterns (DAMPs). These “alarmins” initiate the recruitment of cells of the innate immune system, particularly neutrophils, which initiate alveolar damage via the release of indiscriminate ROS and proteinases, as well as neutrophil extracellular traps (NETs) (Giuranno et al., 2019). In this context, the alarmin, high-mobility group box 1 (HMGB1) and the chemokine, IL-8, are major inducers of NETosis (Tadie et al., 2013; An et al., 2019) with NETs having been described as playing a key role in the “development of ILD of different etiologies” (Porto and Stein, 2016).

Drug-induced interstitial lung disease may manifest as an acute interstitial pneumonitis mediated by cytotoxic and immunological mechanisms (Matsuno, 2012). Reactive oxidants that are toxic to the lung and cytokine-mediated activation of inflammatory cells play important roles in promoting DILD.

Although several clinical risk factors for the development of ICI-Pneumonitis have been identified, little is known about the immunopathogenesis of this condition. Known risk factors include type of malignancy (not surprisingly NSCLC, as well as renal cell carcinoma), bronchial asthma, chronic obstructive pulmonary disease (COPD), and prior radiotherapy (Giuranno et al., 2019). Although not rigorously explored, measurement of systemic biomarkers of sub-clinical inflammation such as the circulating neutrophil count, interleukin (IL)-6, and C-reactive protein (CRP) may help identify those at risk, while, as mentioned earlier, measurement of thyroid and antinuclear autoantibodies may also have predictive potential (Adegunsoye et al., 2015; Adegunsoye et al., 2017; Bonella et al., 2019; Buendía-Roldán et al., 2020).

The most probable mechanism of induction of ICI-Pneumonitis is the recovery of pulmonary T cell reactivity to inhaled antigens, autoantigens, or microbial products following ICI-mediated reversal of the constraints (CTLA-4 and PD-1) imposed on Th1 and Th17 cells by Tregs. In this context, a recent report by Wang et al. is noteworthy (Wang et al., 2020). These authors measured the levels of the counteracting cytokines IL-17A (pro-inflammatory) and IL-35 (immunosuppressive, produced by Tregs) in bronchoalveolar lavage fluid (BALF) from patients (n = 13) with NSCLC who were experiencing episodes of ICI-Pneumonitis. ICI-treated, non-ICI-Pneumonitis NSCLC patients (n = 20) were included as controls. The test cytokines and the percentages of Th1, Th2, Th17, and Tregs were also measured in blood. The authors noted higher frequencies of Th1 and Th17 cells in blood that correlated positively with IL-17A (blood and BALF) and higher ratios of Th17 cells to Tregs during development of ICI-Pneumonitis. Although the numbers of patients recruited to the study were small and given that independent confirmation of these findings is necessary, they nevertheless appear to implicate Th17 cells in the immunopathogenesis of ICI-Pneumonitis, as well as a counteracting role for IL-35 (Wang et al., 2020). Partially confirming these results, another report by Suresh et al. revealed that high lymphocytosis was associated with ICI-Pneumonitis in BALF from ICI-treated patients with lung and skin cancers (Suresh et al., 2019b). While healthy individuals and ICI-treated patients that did not develop ICI-Pneumonitis had 10% lymphocytes and 85% macrophages as predominant BALF immune populations, ICI-Pneumonitis in ICI-treated patients was associated with a lymphocytosis of greater than 20%. To investigate the phenotype of these lymphocytes, the authors performed a comprehensive flow cytometry analysis and found that central memory and TNF-α^highIFN-γ^high CD8 T cells were enhanced in ICI-Pneumonitis patients, in the setting of attenuation of immunosuppressive CTLA4^highPD1^high Treg cells.

Moreover, they also observed an increase in the numbers of IL-1β^hiTNF-α^hiCD-11b^hiCD14^intCD16^int inflammatory myeloid cells (Suresh et al., 2019b). Although this work may not have confirmed a clear role for IL-17A, it became evident that in response to ICI treatment, patients that develop ICI-Pneumonitis recruit increased numbers of inflammatory cells to the lungs, while immunoregulatory populations decreased in number and potency.

Additional mechanisms that may be involved in the immunopathogenesis of ICI-Pneumonitis include: 1) autoantibodies; 2) auto-inflammatory cytokine cascades; 3) neutralization of type M2 macrophages and myeloid-derived suppressor cells; and 4) transition of the pulmonary microbiome to a pro-inflammatory phenotype.

The onset of ICI-Pneumonitis is most often heralded by symptoms of dyspnea (53–79%), dry cough (35–88%), low-grade fever (12%), and chest pain (7%). The time from the administration of the first dose of an ICI to the occurrence of ICI-Pneumonitis varies widely from 1.9 to 24 months, with a median time of 2.8 months (Hassel et al., 2017; Wang et al., 2021). Earlier occurrences have been reported with combination therapies (median 2.7 vs. 4.6 months) and among patients with lung cancer (2.1 months) compared to melanoma (5.1 months) (Zhu et al., 2020).

Pneumonitis terminology is standardized, and associated symptoms are graded on a 5-point scale in accordance with the common terminology criteria for adverse events (CTCAE). Compatible radiographic changes occurring in the absence of symptoms (grade 1) or in the presence of mild symptoms that do not limit normal activities of daily living (grade 2) are designated as mild-to-moderate (low-grade) pneumonitis. High-grade pneumonitis is indicated by severe or medically significant symptoms that are not immediately life-threatening (grade 3), while life-threatening symptoms that require urgent medical attention are considered grade 4 pneumonitis. Pneumonitis-related death is designated as grade 5 (Table 2) (Basch et al., 2014; Kluetz et al., 2016). Mild-to moderate (grade 1–2) pneumonitis accounts for approximately 73–85% of ICI-Pneumonitis across tumor types. Higher rates of severe and fatal pneumonitis have been reported among patients with NSCLC, pre-existing fibrotic lung disease, and among patients treated with dual ICI therapy, ICI plus potentially pneumotoxic chemotherapy combinations and following talc slurry pleurodesis (Suresh et al., 2018b; Sakata et al., 2018; Wang et al., 2018; Moey et al., 2020; Yamagata et al., 2021; Yu et al., 2021).

A diverse spectrum of radiographic features of pneumonitis may be found on CT imaging of the thorax. Five major subtypes have been recognized (Table 3): organizing pneumonia (OP), ground-glass opacities (GGO), nonspecific interstitial pneumonitis (NSIP) (Figure 1), hypersensitivity pneumonitis (HP), and acute interstitial pneumonia (AIP) (Figure 2) associated with the acute respiratory distress syndrome (ARDS) (Nishino et al., 2016b; Naidoo et al., 2017). The radiologic subtypes are consistent throughout the clinical course of ICI-Pneumonitis, although multiple histologic subtypes may coexist in the same patient, while evolution to other histologic subtypes over the course of ICI-Pneumonitis has been reported (Koelzer et al., 2016; Naidoo et al., 2017). Radiographic patterns of pneumonitis may correlate with clinical severity and grades of pneumonitis. AIP and ARDS correlate with more severe pneumonitis and a worse prognosis followed by OP, while NSIP and HP are associated with lower-grade pneumonitis. Thus, radiographic correlates of pneumonitis are significant prognosticators of pneumonitis outcomes and may assist in the management, follow-up, and monitoring of these patients (Nishino et al., 2016a; Nishino et al., 2017b).

Radiation therapy (RT) is a standard treatment modality for certain thoracic malignancies in both the curative and palliative settings. Radiation pneumonitis is an acute manifestation of radiation-induced lung damage and typically occurs 4–12 weeks after treatment is completed. Characteristic radiologic features include ground-glass opacities and airspace consolidation, usually located within or at the edge of the irradiated field. Dyspnea, cough and chest pain are typical symptoms, which can be challenging to distinguish from other comorbidities, such as COPD.

The theory that a synergistic therapeutic effect exists between prior chest RT and ICIs has previously been reported (Herrera et al., 2017). This has led to the implementation of studies that aim to establish the safety and efficacy of concurrent radio-immunotherapy. On the other hand, however, there are studies that have suggested that prior chest RT is a risk factor for the development of ICI-Pneumonitis due to underlying radiation-induced lung damage (Zhai et al., 2020). The exact mechanism underpinning this event remains an area of active interest.

The advent of ICIs has impacted significantly on the outcome of patients with unresectable stage III and metastatic NSCLC in particular. These patients require multimodality treatment and receive RT in either the curative or palliative settings. In this context, Voong et al. (2019) investigated the relationship between RT and ICI-Pneumonitis in patients with advanced NSCLC. Analysis of RT data included intent (curative vs. palliative), dose (Gy), number of fractions, treatment site (chest vs. non-chest), RT technique, and dosimetric parameters (mean lung dose and percentage lung volume receiving 20Gy). Notably, no specific RT treatment parameter was identified that increased the risk for the development of ICI-Pneumonitis. However, on subset analysis, patients who had received curative doses of chest radiation followed by ICI-based therapy were more likely to develop ICI-Pneumonitis (Voong et al., 2019).

The PACIFIC trial (Antonia et al., 2018) investigated the efficacy of definitive chemo-radiation followed by 1 year of consolidation with durvalumab (PD-L1 inhibitor) versus placebo on progression-free survival (PFS) and overall survival (OS) in patients with locally advanced, unresectable NSCLC. Sub-analysis of the prevalence of pneumonitis revealed that while there was a higher rate of all-grade pneumonitis in the durvalumab cohort, no difference was found in the rates of ≥ grade 3 pneumonitis (3.4 vs. 2.6%, no p-value).

In addition, the phase 1 KEYNOTE-001 study investigated the efficacy and safety of pembrolizumab (PD-1 inhibitor) in patients with advanced NSCLC (Garon et al., 2019). A secondary analysis of the data from this study, undertaken by Shaverdian et al. (2017), revealed that the duration of OS was longer (10.7 vs. 5.3 months) in patients who had previously received RT. Of the 24 patients who had previously received thoracic RT, 63% developed pulmonary toxicity versus 40% in the cohort who had not. Notably, there was no difference between the two groups with respect to the number of patients who developed ≥ grade 3 pulmonary toxicity (Shaverdian et al., 2017).

In patients receiving combined modality treatment, it becomes challenging for clinicians to differentiate whether pneumonitis is due to immunotherapy, RT, or a combination of the two. A study by Suresh^b et al. designed to investigate this issue established that the median time to ICI-Pneumonitis development was 82 days after the initiation of ICI, which is similar to the onset of radiation pneumonitis (Suresh et al., 2018b). In addition, the overlapping symptoms and radiographic features of radiation pneumonitis and ICI-Pneumonitis can make it difficult for clinicians to distinguish which type of pneumonitis a patient has developed. In this context, it is noteworthy that radiation pneumonitis usually occurs in, or in close proximity to, the irradiated field, while ICI-Pneumonitis most commonly develops at the edge of the radiation field or in a non-irradiated region (Voong et al., 2019). It is crucial to consider the effects of prior radiation therapy when evaluating a patient for suspected ICI-Pneumonitis.

Establishment of temporal eligibility (symptom development following initiation of the drug) and knowledge of typical symptoms and the usual latency period following drug exposure help to raise suspicion of ICI-Pneumonitis. Multi-disciplinary collaborations involving physicians with expertise in pulmonary medicine and infectious diseases and radiologists may expedite diagnosis and facilitate uniform management strategies. Diagnostic confirmation requires the exclusion of competing disease entities such as disease progression, infection, or thromboembolism. Thoracic computed tomography permits the identification of radiographic patterns of lung abnormalities and is the preferred imaging strategy. Analysis of blood, sputum, urine, nasal swabs and bronchoscopically-obtained lavage fluid may help to exclude infection. Evaluation for infection may be tailored to the index of suspicion for infection and the grade of suspected pneumonitis at presentation (Suresh et al., 2018a). Multiple society guidelines now advocate bronchoscopy with bronchoalveolar lavage and a complete infectious panel to evaluate patients with pneumonitis grades 2 or higher prior to initiation of immunosuppressive treatment. Bronchoalveolar lavage (BAL) fluid specimens typically demonstrate a lymphocytic pleocytosis (Suresh et al., 2019a). Lung biopsies are not a usual requirement for the diagnosis of ICI-Pneumonitis, but may assist in excluding other diagnoses, such as tumor progression. Histologic descriptions of ICI-Pneumonitis range from NSIP, OP, granulomatous inflammation, and cellular interstitial pneumonitis. Less commonly, acute fibrinous OP and diffuse alveolar damage are reported (Naidoo et al., 2017; Wang et al., 2021). Associated inflammatory infiltrates are typically lymphocyte-predominant. Poorly formed granulomas and increased numbers of eosinophils may also be seen (Naidoo et al., 2017; Hattori et al., 2019; Jodai et al., 2019). Increased levels of interleukin-6 (IL-6), tumor necrosis factor-α (TNF-α), interleukin-17 (IL-17), and to a lesser extent, interleukin-8 (IL-8) and interleukin-10 (IL-10) represent nonspecific signals of inflammation, but cannot distinguish inflammation in the setting of infection from a drug-related inflammatory response (Larsen et al., 2019; Wang et al., 2021). Differential expression of IL-17 in BAL IFNγ⁺ IL-17^- CD8⁺ T and CXCR3⁺ CCR6⁺ Th17/Th1 cells was noted in BAL fluid of patients with ICI-Pneumonitis compared to patients with pneumonia; however, larger studies are needed to confirm these findings (Kim et al., 2021). If available, changes in pulmonary function tests (PFTs) and 6-min walk tests (6 MWT) from baseline may be helpful in detecting lung function changes in ICI-treated patients. In addition, these studies may assist in risk stratification of patients with pre-existing lung disease and determine the severity of pneumonitis, while guiding response to specific ICI-Pneumonitis therapy. These tests have been incorporated into multiple national guidelines as simple, inexpensive, and non-invasive monitoring tools that can be performed at baseline before initiating ICI-based therapies (Brahmer et al., 2021). However, as demonstrated with other forms of drug-induced pneumonitis, the sensitivity of PFTs and 6 MWT in the early detection of ICI-Pneumonitis has not been established.

In conjunction with corticosteroid therapy, it is essential that the patient receives appropriately targeted oxygen therapy and chest physiotherapy to support good sputum drainage. Judicious empirical antimicrobial therapy in accordance with local guidelines should be commenced while infection screens are being processed. Early decisions regarding escalations of care, including whether intubation and ventilation should be undertaken.

Other rare immune-mediated respiratory emergencies include myositis of the diaphragm, which presents with rapidly progressive neuromuscular hypoventilation (Haddox et al., 2017; John et al., 2017). In accordance with the management of other life-threatening, immune-mediated toxicities, these patients require early recognition with the initiation of high-dose corticosteroids, early immunosuppression, and critical care support.

Pulmonary embolism (PE) is more common in melanoma patients treated with ICIs (Sussman et al., 2021). Furthermore, it appears to be more prevalent in patients treated with combination checkpoint inhibition in comparison with single-agent therapy. A high clinical suspicion for PE is thus required in dyspnoeic patients with no other apparent cause, and therapeutic anticoagulation should be commenced after confirmation of diagnosis.

Rare pulmonary complications associated with nivolumab include acute pulmonary hypertension and an increased pulmonary artery diameter (Fournel et al., 2020).

COVID-19 may mimic presentations of ICI- Pneumonitis and is a recent important addition to the differential diagnosis in patients with acute respiratory symptoms treated with ICI therapy (Cooksley et al., 2021).

The COVID-19 pandemic caused by severe acute respiratory syndrome coronavirus 2 (SARs-COV-2) has disproportionately affected patients with cancer (Bakouny et al., 2020; Garassino et al., 2020). Weakened T cell activity caused by exhaustion of T lymphocytes is a distinguishing feature of cancer and chronic infections. Characteristic findings of exhausted T-lymphocytes include overexpression of PD-1 and CTLA-4 immune checkpoint receptors and loss of IL-2 production, impaired cytotoxicity, diminished proliferation, and altered production of proinflammatory cytokines. T cell exhaustion is a hallmark of COVID-19 infection (Dyck and Mills, 2017; Hashimoto et al., 2018; Wykes and Lewin, 2018). ICIs can abrogate T cell exhaustion and depletion by blocking CTLA-4 and PD-1 inhibitory signaling. These observations suggest that cancer patients with concomitant infection might derive benefit from ICI therapy.

On the other hand, probable upregulation of hitherto suppressed positive immune checkpoint molecules in effector T cells and resultant generalized inflammatory cytokine secretion may exacerbate end-organ damage associated with cytokine release syndrome, leading to more severe COVID-19 symptoms. These observations have raised concerns regarding the safety of ICI therapy for cancer patients in the setting of COVID-19. The TERAVOLT study, a multinational investigation of 200 patients with thoracic cancer and COVID-19, reported that the mortality associated with COVID-19 infection in patients with active thoracic cancers was 33%. The type of systemic therapy (tyrosine kinase inhibitor, conventional chemotherapy, or ICI therapy) did not impact survival, suggesting that withholding or discontinuing ICI therapy for cancer patients with concomitant COVID-19 might not be warranted (Garassino et al., 2020). Several other smaller studies corroborate these findings (Kattan et al., 2020; Luo et al., 2020; Quaglino et al., 2020). These results must be considered with caution, and further prospective investigations are needed.

Cytokine release syndrome (CRS) is very rarely observed following ICI therapy and appears clinically indistinguishable from a cytokine storm, a key feature of advanced COVID-19. However, overlapping clinical features of ICI- Pneumonitis, which is much more common, and COVID-19 pneumonitis create a more frequent conundrum, rendering accurate diagnosis and appropriate management difficult. Systemic glucocorticoids are the mainstay of treatment for grade 2 or higher ICI-Pneumonitis and during the early stages of COVID-19 pneumonia with ARDS. Cytokine excess, in particular IL-2, IL-4, IL-6, TNF-α, and IFN-Ɣ are known drivers of the overactive inflammatory response and associated COVID-19 severity (Liu et al., 2020; Du et al., 2021). The TNF-α inhibitor, infliximab, and the IL-6 inhibitor, tocilizumab, are used as second-line therapies in patients with steroid-refractory irAEs, including ICI-Pneumonitis. Their role in the management of COVID-19 is under intense investigation, with early evidence from the REMAP-CAP trial showing therapeutic promise (Gordon et al., 2021).

Both CTLA-4 and PD-1/PD-L1 inhibitors have been added to a growing list of agents implicated in developing granulomatous syndromes indistinguishable from sarcoidosis (Vogel et al., 2012; Kim et al., 2016). Although mechanistic underpinnings of ICI-associated sarcoid-like reactions have not been fully elucidated, amplification of Th17 by CTLA-4 and PD-1 axis blockade is thought to play an integral role (Vogel et al., 2012; Huang et al., 2013; Braun et al., 2014; Broos et al., 2015). Noncaseating granulomas, the cardinal feature of sarcoid-like reactions, are indistinguishable from confirmed sarcoidosis. Sarcoid-like reactions most commonly involve the lungs, lymph nodes, and skin. Other extrathoracic manifestations involving the parotids, spleen, bone, eyes, and brain have also been reported. Patients typically present at 12 weeks (range 3 weeks to 2 years) following drug administration with asymptomatic mediastinal and/or hilar lymphadenopathy and upper lobe predominant pulmonary nodules in a perilymphatic distribution on CT scan (Figure 3) (Kim et al., 2016; Rambhia et al., 2019). Dry cough and dyspnea are occasionally reported. No apparent ICI dose threshold that triggers sarcoid-like reactions has been identified. ICI-related sarcoid-like reactions show fluorodeoxyglucose uptake (FDG) in involved tissues on positron-emission tomography (PET) scans that are indistinguishable from confirmed sarcoidosis and sarcoid-like reactions from other causes of lymphadenopathy, including malignancy. Hypercalcemia and elevations in serum angiotensin-converting enzyme levels have also been observed (Kim et al., 2016). Pulmonary function studies are typically normal. BAL fluid commonly demonstrates a CD4-rich lymphocytic alveolitis. The diagnosis requires examination of tissue samples obtained by bronchial biopsy, transbronchial lung biopsy, or biopsies of the mediastinal lymph nodes using endobronchial ultrasound (Luke et al., 2015). Findings of focal infiltration by noncaseating epithelioid and giant cell granulomas establish the diagnosis once other causes have been excluded. In most patients, resolution of signs and symptoms following drug-interruption occurs with or without the initiation of steroids. Spontaneous resolution in early disease despite continuation of ICIs has also been reported. Recurrence of sarcoid-like reactions following resumption of ICI therapy may occur. The indication, dose, and duration, and optimal tapering schedule of steroid therapy in this context has not been defined (Berthod et al., 2012; Murphy et al., 2014; Danlos et al., 2016).

Specific drugs may trigger radiation injury within a previously irradiated field. This phenomenon, known as radiation recall pneumonitis (RRP), may be provoked by a variety of agents, including the taxanes, anthracyclines, gemcitabine, erlotinib, etoposide, oxaliplatin, and vinorelbine (Burris and Hurtig, 2010; Ding et al., 2011). More recently, RRP has been reported following anti-PD1 therapy (Gupta and Lee, 2017; Shibaki et al., 2017). In one study, the incidence of PD-1-triggered RRP among patients with NSCLC was 4.2% but rose to nearly 31% among patients with a prior history of PD-1 related pneumonitis (Gupta and Lee, 2017). The development of new ground-glass and consolidated opacities within the irradiated field signals RRP development, which typically appears 3 months to 2 years after completion of radiation therapy and one to 2 months after exposure to the triggering chemotherapeutic agent (Gupta and Lee, 2017; Shibaki et al., 2017). The pulmonary infiltrates may be asymptomatic or associated with dry cough, dyspnea, chest pain, and low-grade fevers (Figure 4). Signs and symptoms typically respond to systemic corticosteroids. We use 0.5–1 mg/kg of methylprednisolone or equivalent with a 4–6-weeks taper, although the optimal dose and duration of steroid therapy have not been established in randomized trials.

Pseudoprogression represents an unconventional clinical response to ICI therapy in which there is an initial increased size of tumor lesions or the appearance of new lesions on CT, with subsequent reduction in tumor burden. This clinical response typically occurs during the first cycles of therapy and has been observed in 2.8–15.8% of all ICI-treated patients across tumor types. Pseudoprogression may be erroneously classified as progressive disease according to the size-based WHO or RECIST criteria, leading to premature discontinuation of the drug. Distinguishing pseudoprogression from true tumor progression based on radiological evidence alone is challenging, particularly early in the course of ICI therapy. Ideally, pathologic findings of peritumoral necrosis or inflammatory cell infiltrates are needed for confirmation; however, tissue biopsies may not be pragmatic in some patients. Thus, the diagnosis of pseudoprogression is based on CT findings of tumor enlargement following the initial cycle of ICI blockage and subsequent shrinkage in tumor burden on subsequent CT, performed at least 4 weeks later. If tumor shrinkage is noted on the subsequent CT, ICI therapy may be safely continued. Shrinkage of tumor elsewhere and the absence of clinical deterioration in the patient’s condition supports the diagnosis of pseudoprogression.

Unlike pseudoprogression, in which there is a radiographically apparent increase in tumor burden after ICI therapy followed by tumor regression, hyperprogression, represents true tumor progression in which there is a very rapid and sustained progression of tumor following the initiation of immunotherapy. This ICI-related rapid surge in tumor burden has primarily been reported following PD-1/PD-L1 therapies for lung cancer (Kato et al., 2000; Chubachi et al., 2016; Kurzrock and Kato, 2018; Champiat et al., 2019).

Severe isolated airway diseases, including bronchitis and bronchiolitis, have been attributed to ICI therapy in individual case reports. Response to bronchodilator therapy and inhaled steroids has been variable (Mitropoulou et al., 2020). Pleural effusion has also been rarely described following PD-1 and PD-L1 therapies (Kolla and Patel, 2016; Ikematsu et al., 2019). Effusions are lymphocyte-predominant and may spontaneously remit despite continuation of ICI therapy.