Immune checkpoint inhibitor (ICI) therapy represents a paradigm shift in the treatment of patients with locally advanced and metastatic lung cancer. Although immunotherapy generally has a more favorable safety profile when compared with chemotherapy, immune-related adverse events represent important, but incompletely understood, treatment-limiting complications associated with significant morbidity and mortality risk. Current guidelines for diagnosis and management are derived from consensus experience, highlighting that further prospective investigation in this area is needed. As ICI-related pneumonitis is a clinically and radiographically diverse toxidrome, clinical vigilance is important while treating patients with lung cancer.
Immune checkpoint inhibitor (ICI)-related pneumonitis is an infrequent but clinically significant consequence of immunotherapy treatment in lung cancer.
Clinicians must maintain a high index of suspicion for ICI-related pneumonitis given the variability in clinical, radiographic, and pathologic presentations and limited understanding of toxicity risk factors.
With prompt diagnosis and intervention, typically with drug cessation and immunosuppression, outcomes after ICI-pneumonitis are generally favorable, and pneumonitis treatment does not appear to impact antitumor response.
Heterogeneity in currently available guidelines for ICI-pneumonitis diagnosis and treatment reflects the variability in clinical experience and highlights the need for future prospective study.
Rechallenge with ICI immunotherapy agents after ICI-related pneumonitis remains a controversial subject for which shared decision-making is essential until there is improved understanding of individual risk factors for recurrent toxicity.
After multiple recent successful treatment trials in lung cancer, immune checkpoint inhibitor (ICI) use is rapidly expanding as a paradigm shift in the treatment of advanced disease. Pulmonary complications with these agents, although relatively infrequent, represent a significant source of morbidity and mortality for patients treated with ICIs. Prompt recognition is essential to appropriate evaluation and treatment, but can be challenged by the variable and nonspecific symptoms, timing of onset, radiographic appearance, and pathology seen with this toxicity. Furthermore, the available guidelines for evaluation and management of patients with ICI-pneumonitis are derived from limited and largely retrospective data that do not necessarily address the diverse patient populations being treated with ICIs today. All clinicians caring for patients with lung cancer must maintain a high index of suspicion and an up-to-date understanding of this unique toxidrome and current principles of management. It is equally important that we acknowledge and understand multiple areas of uncertainty in this field that require future experience and investigation.
In the 2 years since the last Clinics review on this topic, the clinical indications for use of ICIs in lung cancer have expanded, and the use of these therapies has moved from academic research centers to the community. As such, our understanding of ICI-related pneumonitis development, risk factors, and management has continued to evolve, and new areas of uncertainty have emerged. The principal goals of this article are (1) to review current state-of-the-art with regard to epidemiology, presentation, evaluation, and management of ICI-related pulmonary complications; and (2) to discuss notable challenges and controversies, prioritizing newest developments since the last review. Although we focus on ICI-related pneumonitis in patients with lung cancer, we acknowledge that many studies in this area include patients with multiple cancer types. Our understanding of these complications is enriched by experiences across these different patient populations. For complete updates on immunotherapy use in lung cancer, we direct the readers to 2 articles elsewhere in this issue of Clinics focused on lung cancer treatment, including Advances in the Treatment of Non–small cell lung cancer (NSCLC): Immunotherapy and Advances in the Treatment of Small Cell Lung Cancer (SCLC).
Where are we now?
Discussion of ICI-related pneumonitis merits brief discussion of the mechanisms of immunotherapy agents and an updated perspective on recent lung cancer treatment trials that guide the role for ICI agents. Cancer immunotherapy is a general term for all therapies that harness the immune system for tumor cell recognition and elimination, and are an outgrowth of innate immune responses to malignancy. However, immune system activity can in turn be downregulated by tumor surface molecules and the tumor microenvironment, allowing tumors to evade detection by the immune system. , Prominent among cancer immunotherapies have been the development of drugs restoring and enhancing the critical immune checkpoint pathways that can be subverted by malignancies.
Two of the best understood checkpoint molecules are cytotoxic T-lymphocyte associated protein 4 (CTLA-4) and programmed cell death 1 protein (PD-1). Both CTLA-4 and PD-1 molecules are receptors on the surface of T cells, and their activation inhibits T-cell function. Interestingly, cancer models have shown that tumor injury induces expression of tumor surface signaling molecules that may facilitate immunoevasion. , For example, in separate models, radiation and chemotherapeutic agents resulted in increased expression of the PD-1 ligand (PD-L1). , Conversely, preventing T-cell inhibition by tumors and/or the local microenvironment activates the antitumor immune response. ICIs are monoclonal antibodies targeting the CTLA-4, PD-1, or the associated ligands (eg, PD-L1).
Just as immune checkpoints represent a critical point of homeostatic control, ICIs have the potential to unleash both positive (antitumor) immune responses and negative (antiself) consequences in an unpredictable fashion. Immune-related adverse events (irAEs) have been observed in essentially every organ system and share similarities to many autoimmune diseases. The precise pathophysiology of irAEs remains opaque and likely involves multiple mechanisms beyond T-cell activity alone.
Since the last Clinics review on this topic, multiple studies have shown improved clinical outcomes in lung cancer using ICIs in a variety of patient populations. Although original work initially supported use of nivolumab as a second-line agent in metastatic NSCLC, , ICIs are now being used as first-line or adjuvant treatments for patients with advanced NSCLC, locally advanced NSCLC, and SCLC. In unresectable stage III NSCLC, Antonia and colleagues reported improved progression-free survival (PFS) and overall survival (OS) by adding durvalumab to conventional chemoradiation regimens. In 2018, pembrolizumab was shown to improve OS in both metastatic squamous and nonsquamous NSCLC when used in combination with chemotherapy. In addition, use of atezolizumab has resulted in improved OS in both nonsquamous NSCLC (in combination with chemotherapy and bevacizumab) and extensive-stage SCLC (in combination with chemotherapy). Finally, in patients with stage IV or recurrent NSCLC with high tumor mutational burden, Hellman and colleagues demonstrated improved PFS with use of combined nivolumab/ipilimumab therapy. As such, most patients with advanced-stage lung cancer without targetable mutations will receive ICI-containing treatment regimens as part of their first-line therapy. Therefore, the number of patients who will develop ICI-related complications (particularly pneumonitis) will almost certainly grow.
Epidemiology: an evolving and complex issue
Our understanding and recognition of ICI-related pneumonitis is still rapidly evolving. As of our last article in 2017, ICI-pneumonitis was considered to be relatively uncommon and generally mild in severity. Although still a relatively infrequent complication of immunotherapy, ICI-pneumonitis is increasingly recognized as a serious, often treatment-limiting complication of therapy. A recent systematic review and meta-analysis reported that ICI-pneumonitis is the most common cause of immunotherapy treatment–related death, with up to 24% of pneumonitis cases grade 3 or higher in severity. O’Connor and colleagues studied a cohort of patients across academic and community oncology practices. They noted that within 4 months of US Food and Drug Administration approval, more than 60% of eligible patients with melanoma, renal cell carcinoma (RCC), and NSCLC received anti-PD-1 treatment. Importantly, compared with patients in ICI clinical trials, the patients in the studied cohort were more likely to be older than 65 years and may have been less proactively screened for comorbid conditions. Such observations have raised concerns that real-world experience with ICI-related complications may diverge significantly from that observed in clinical trial cohorts.
Indeed, subsequent studies have shown variable incidence of ICI-related pneumonitis compared with earlier studies. The 2016 systematic review and meta-analysis of PD-1 inhibitor use in patients with melanoma, RCC, and NSCLC by Nishino and colleagues reported an overall incidence rate of 2.7%; incidence of high-grade disease (ie, grade 3 or higher) was 0.8%. Similarly, in 917 patients (multiple tumor types) who received anti-PD-1 or anti-PD-L1 therapies (hereafter referred to as anti-PD-[L]1), Naidoo and colleagues reported an overall incidence of 5%, although most of this represented grade 1 or 2 disease. In contrast, in a single-center analysis of both clinical trial and nontrial patients with lung cancer (n = 205), Suresh and colleagues observed a much higher incidence of ICI-pneumonitis (19%) than in prior studies. Although the most recent work needs multicenter validation, it seems likely that ICI-pneumonitis risk and outcomes will not be equivalent between clinical trial and non–clinical trial patients, and therefore true epidemiology remains yet to be determined.
Several potential risk factors for developing ICI-related pneumonitis have been explored, including tumor-related factors, patient-related factors (eg, exposures or comorbid diseases), and treatment-related factors (eg, ICI class or use in combined modality regimens). Starting with tumor-related risk factors, the systematic review by Khoja and colleagues (n = 48 trials and 6938 patients) of immunotherapy (anti-CTLA-4, anti-PD-1, anti-PD-L1, and anti-CTLA-4/PD-1) reported higher rates of pneumonitis in NSCLC and RCC (compared with melanoma). In their systematic review and meta-analysis (n = 26 articles and 4496 patients), Nishino and colleagues also reported higher risk of pneumonitis related to PD-1 therapy in patients with NSCLC. Finally, Suresh and colleagues reported increased risk of pneumonitis in patients with squamous cell carcinoma compared with other histologies. Although the mechanisms behind an individual’s sensitivity to ICI-related pneumonitis are not yet clear, tumor types and histology appear to play a role. Although there has not been a clear association between PD-L1 status and ICI-related toxicity risk, tumor mutational burden may be emerging as a risk factor worth further consideration. Tumor mutational burden has already demonstrated an association with improved treatment outcomes and durable treatment responses in a fashion independent of PD-L1 status. In addition, however, patients with high mutational burden appear to be at substantially increased risk for development of ICI-related irAEs. Indeed, it has been postulated that up to 50% of the differences observed in irAE incidence may be due to tumor mutational burden.
Turning to patient-related risk factors, several investigators have identified increased risk of ICI-pneumonitis related to smoking patterns or preexisting lung disease. Naidoo and colleagues identified PD-(L)1-related pneumonitis in both current/former smokers and never-smokers, although more patients with pneumonitis were current/former smokers (56% vs 44%). In general, clinical outcomes were worse in current/former smokers and patients with underlying lung disease, although this may reflect the burden of underlying pulmonary comorbidities as well. In 2018, Yamaguchi and colleagues reported higher rates of PD-1-related pneumonitis in patients with NSCLC and underlying pulmonary fibrosis. In those patients with NSCLC undergoing treatment with nivolumab or pembrolizumab, 35.1% of those with baseline pulmonary fibrosis developed ICI-pneumonitis (compared with 5.8% of those without fibrosis). However, it is not uniformly true that patients with underlying interstitial lung disease go on to develop ICI-related pneumonitis, as other small case series have demonstrated successful treatment of patients with underlying interstitial lung diseases. ,
The clinical implications of using ICIs in patients with preexisting autoimmune disease remain difficult to quantify across all types of irAE; theoretic risks must be balanced against the potential consequences of withholding potential life-saving therapies in these populations. Leonardi and colleagues evaluated a cohort of patients with underlying autoimmune diseases treated with ICIs (n = 46), although notably none in this cohort had interstitial lung disease. Most of these patients tolerated ICI immunotherapy well, and those who developed either a flare of their underlying autoimmune disease or an unrelated irAE improved promptly with cessation of therapy; a minority of patients required additional corticosteroid therapy. Notably, of the high-grade irAE observed, a third of the cases were ICI-related pneumonitis, all of which were ultimately steroid responsive. Although these findings do suggest that assessment of preexisting autoimmune disease and discussion of potential risks should be part of the shared decision-making surrounding ICI therapy, presence of such conditions should not automatically preclude consideration of ICI therapy. Unfortunately, we currently have no biomarkers or clinical prediction tools that differentiate those patients with underlying autoimmune diseases who may be at highest risk, although this same study suggested that those patients who had clinically active disease at time of ICI initiation were at higher risk for complications.
Regarding treatment-related risks, it has already been established that individual ICI agents and classes appear to confer different risks of adverse events (although not specifically pneumonitis, or even irAE-specific complications). Two recent systematic reviews and meta-analyses are instructive in this regard. , Wang and colleagues evaluated 125 clinical trials of single-agent PD-(L)1 inhibitors and reported higher incidence of all-grade adverse events from nivolumab (compared with pembrolizumab); PD-1 inhibitors also had a higher incidence of high-grade adverse events compared with PD-L1 inhibitors. The meta-analysis by Xu and colleagues included randomized trial of multiple immunotherapy agents (ie, nivolumab, pembrolizumab, ipilimumab, tremelimumab, and atezolizumab), conventional chemotherapy, and combination therapies. Overall, atezolizumab had the most favorable safety profile, although nivolumab appeared safer in patients with lung cancer. Importantly, these findings referred to all adverse events and not specifically immune-related adverse events, therefore generalizability of these data are limited.
In general, ICI-related pneumonitis is more commonly seen in PD-(L)1 or combination PD-(L)1/CTLA-4 therapies compared with CTLA-4 agents alone. Khoja and colleagues also found higher risk of pneumonitis from PD-1 agents compared with CTLA-4 agents; there was no dose-dependent effect. In a systematic review and meta-analysis (n = 19 trials), Khunger and colleagues noted higher risk of ICI-related pneumonitis with PD-1 agents compared with PD-L1 agents, with treatment-naïve patients at higher risk than previously treated patients, perhaps due to the immunosuppressive effects of prior therapy. Studies by Nishino and colleagues and Naidoo and colleagues noted that pneumonitis risk was particularly high in patients receiving combination immunotherapy (ie, PD-1/CTLA-4) compared with those receiving single-agent immunotherapy; this has been generally true of all irAEs. Finally, at least 2 studies have shown significantly increased risk of pneumonitis in NSCLC when immunotherapy is used in association with epidermal growth factor tyrosine kinase inhibitors (EGFR-TKIs). , In a retrospective database review of 20,516 patients with NSCLC, Oshima and colleagues showed a high proportion of pneumonitis in patients with NSCLC receiving both nivolumab and EGFR-TKI therapy (26%). Subsequently, in a single-center analysis (n = 126 patients with EGFR mutant NSCLC treated with EGFR-TKIs and immunotherapy), Schoenfeld and colleagues reported development of severe ICI-pneumonitis in 15% of cases when PD-1 or PD-L1 therapy was followed by osimertinib.
The consequences of combined radiation therapy and ICI therapy remain challenging to parse, in large part due to the overlapping timing, clinical presentation, and radiographic appearance of radiation-induced lung injury and ICI-related pneumonitis. However, patients with locally advanced lung cancer who received chemoradiation in combination with durvalumab had a higher rate of pneumonitis (12.6%) than those who received chemoradiation alone (7.7%). Cases of “radiation-recall” have been described during ICI therapy. , Although several radiation treatment–related risk factors have not yet been explored, there is the suggestion that curative intent therapy may carry a higher risk for synergistic ICI-related pneumonitis than palliative radiation. In summary, there are likely multiple patient, tumor, and treatment regimen factors that require further exploration with regard to the risk of ICI-related pneumonitis.
The “typical” presentation of immune checkpoint inhibitor–related pneumonitis
Variability remains the hallmark feature of ICI-pneumonitis. Although most cases emerge within weeks to months of immunotherapy initiation, pulmonary toxicity has been observed as early as after the first dose, and as late as months or years into therapy. , Furthermore, relapses of ICI-pneumonitis have been observed after drug cessation, extending the window of suspicion even longer.
As with most drug-induced lung disease, clinical symptoms are nonspecific. Nonproductive cough and progressive dyspnea are the most common symptoms observed, although fever and chest pain also may be reported. , Because many patients with lung cancer have significant underlying pulmonary comorbidities, deviations from symptom baseline can be difficult to discern and require a high degree of suspicion. Up to a third of patients with mild ICI-related pneumonitis may be asymptomatic at presentation.
Physical examination and laboratory findings are likewise nonspecific; in contrast to other irAEs such as thyroiditis, there are currently no serologic biomarkers that are sensitive or specific for ICI-pneumonitis, although this remains an area of active investigation. However, coexistence of symptomatic or subclinical extrapulmonary irAE is relatively common in individuals with ICI-related pneumonitis, so the presence of an extrapulmonary irAE should therefore raise the index of suspicion in patients with new pulmonary symptoms.
Chest radiographs are notoriously insensitive for detection of ICI-related pneumonitis, particularly in milder cases. Moreover, findings related to the underlying malignancy or to comorbid pulmonary conditions can further complicate and confound radiographic interpretation. Although focal “organizing-pneumonia”–type infiltrates ( Fig. 1 ) and ground-glass opacities ( Fig. 2 ) are the most common radiographic patterns observed, no one pattern is pathognomonic for ICI-toxicity. Computed tomography (CT) scans may demonstrate a wide spectrum of abnormalities, including diffuse ground-glass infiltrates, focal consolidation, interlobular septal thickening (with or without reticular changes and/or honeycombing), pulmonary nodules (suggesting hypersensitivity pneumonitis-type pattern or sarcoidlike reaction, Fig. 3 ), or mixed patterns including more than 1 of those listed previously. , Figs. 1–3 provide examples of representative cases.