Lung cancer has the greatest death rate, at 25%, of all types of cancer, with an estimated 135,720 deaths in the United States in 2020 (1). Non–small cell lung cancer (NSCLC) is the most common lung cancer subtype, and it comprises two major histological types: squamous cell carcinoma (SCC) and adenocarcinoma (AC) (2). Nearly 70% of patients with NSCLC are initially diagnosed at a locally advanced stage and suffer from a poor prognosis (2). The 5-year survival rate is less than 3% for patients with advanced NSCLC (3). Historically, the standard management recommended for patients with NSCLC who present with advanced-stage disease was chemotherapy regimens combined with radiotherapy (RT). However, the treatment provided generally modest responses, with an overall survival (OS) of approximately 12 to 18 months and a median progression-free survival (PFS) of just 4 to 8 months (4, 5).

Recently, immunotherapy that includes programmed cell death-1 (PD-1), programmed cell death ligand 1 (PD-L1), and cytotoxic T-lymphocyte antigen 4 (CTLA-4) inhibitors, which enhance anti-tumor activity, has revolutionized the therapeutic strategy for multiple malignancies (6). PD-1, a type I transmembrane protein, exists inherently on activated T cells, B cells, natural killer cells, macrophages, dendritic cells, and monocytes. PD-L1 is highly expressed on both cancer cells and antigen-presenting cells (7). The interaction of these two molecules could promote self-tolerance and attenuate autoimmunity through T-cell exhaustion and reduced cytokine production (8). CTLA-4, a critical surface protein receptor and co-inhibitor, is typically located in stimulated CD4+/CD8+ T cells to dampen T-cell activity by binding CD80/CD86/CD28. Using the inhibitory mechanism checkpoint pathways or molecules, immune checkpoint inhibitors (ICIs) can tilt the immune equilibrium toward the beneficial promotion of tumor killing and the boosting of an immune attack (6, 9).

In advanced NSCLC, an increasing body of clinical studies suggests that the application of ICIs could achieve significant breakthroughs in PFS and OS (10–13). Therefore, the US Food and Drug Administration has rapidly incorporated ICIs into first-line therapies for advanced NSCLC (14). In the PACIFIC regimen, durvalumab (a PD-L1 inhibitor) has become the new standard of care after platinum-based chemoradiotherapy for unresectable stage III NSCLC in the United States, Europe, and Japan (15).

However, along with the killed tumor cells, virtually every organ system could be affected by ICIs (5). Immune-related adverse events (irAEs), such as cutaneous lesions, myocarditis, hepatitis, colitis, endocrinopathies, inflammatory arthritis, and pneumonitis, are widely reported (15, 16). The incidence of irAEs might be higher with combination ICI use, specific cancer types, and non-trial conditions (17, 18). Among all reported irAEs, checkpoint inhibitor pneumonitis (CIP) is particularly worrisome and potentially lethal (18–21). CIP may occur more often and have a faster onset in NSCLC than in other types of cancer (22). Since before ICI therapies, pulmonary function has been compromised by tumor location and size in patients with NSCLC. In addition, pre-existing lung comorbidities, such as chronic inflammatory respiratory diseases, interstitial fibrosis lung diseases, and radiation-induced pneumonitis (RIP), may cloud diagnostic accuracy because of the overlapping respiratory symptoms and signs (5, 6, 9, 14, 23). As a result, recognizing the unique clinical and imaging patterns of CIP is essential to facilitate expeditious diagnosis and optimized management principles.

Although previous studies have elucidated the incidence, potential mechanisms, diagnosis, risk factors, and management of CIP, they focused on variable focuses that were not comprehensive and deep enough (5, 6, 9, 14, 23, 24). This review offers a summary of cases or case series concerning CIP in NSCLC, and it aims to identify the characteristics of typical patients who develop CIP. We also comprehensively summarize the current knowledge and relevant studies of ICI-associated pneumonitis, and we discuss the future direction of evidence-based therapeutic schedules for patients with CIP in NSCLC.

The definition of CIP is the occurrence of respiratory symptoms/signs related to a new emerging infiltration viewed on a chest X-ray but excluding new infections tested by sputum and/or bronchoalveolar lavage (BAL) (5). In different tumor types, the overall incidence of CIP varied from 3% to 5% for all grades and ranged from 0.8% to 1.0% for grade ≥ 3 CIP (5, 14, 25, 26). The overall fatality rate of CIP was 10% to 17%. In NSCLC, the incidence of CIP mainly originated from clinical trial and real-world data. In clinical trial data (10, 27–44), the incidence of CIP for all grades was approximately 2% to 38%, and incidence for grade ≥ 3 CIP was approximately 0.6% to 2.7%. In real-world data, the incidence of CIP in patients with NSCLC was 4.8% to 39.3% (18, 24, 27, 28, 45–52). The discrepancy between data from these two sources might be partly attributed to the increasing awareness of CIP in the medical community, which contributed to more frequent clinical detection and less stringent inclusion criteria for real-world studies compared with randomized trials.

The median time to the onset of CIP was typically approximately 2.8 months, and the overall range spanned from 9 days to 19.2 months (18, 20, 53, 54). We included 44 occurrences of CIP in patients with NSCLC ( Figure 1 ; Table 1 ) by searching Pubmed and Web of Science from 2016 up to April 15^th, 2020. We used the search terms “immune checkpoint inhibitors *” OR “immunotherapy *” AND “non-small cell lung cancer*” AND “pneumonitis*” with related terms including MeSH terms as well as keywords. All case reports were included. And we found that the mean time to CIP onset from the start of ICI therapy was approximately 10 weeks (2.5 months; Table 2 ). No difference was found in the median time from treatment to CIP onset between patients with improved/resolved CIP and deteriorated/maintained CIP (P=0.547) ( Table 2 ). The onset of CIP reportedly occurred as early as hours to days—or as late as several months—after the first ICI dose; however, more severe CIP grades usually had onset within the first 100 to 200 days of ICI therapy (87). The median time to CIP onset was not related to disease severity (88), and onset seemed to occur earlier for patients treated with combination ICIs (18). Of note, CIP might develop months after therapy termination, which suggests that continuous vigilance after drug discontinuation is necessary (54).

In animal models with deficiencies of PD-1 and CTLA-4, animals exhibited lung infiltration (89, 90), which could clarify questions about how CIP develops (91). The potential mechanisms driving ICI-related pneumonitis are outlined in the following sections.

Current evidence from retrospective studies and case reports has identified many potential risk factors for ICI-related pneumonitis (6, 24, 53, 72, 103–105). These include baseline patient characteristics, disease features, and therapy management. Specific factors include age, sex, smoking status, previous lung disease, tumor histological type, PD-1 blockade, combination therapy, and prior RT.

Because specific clinical or radiologic markers are absent, diagnosis of CIP is quite difficult. CIP is typically a diagnosis of exclusion, one that should rule out infection, tumor progression, and radiation-related pneumonitis (25). New emerging or the deterioration of respiratory symptoms—especially dry cough, dyspnea, and decreasing oxygen saturation—after ICI therapy for NSCLC require consideration of CIP (23). The diagnostic workup ( Figure 3 ) to identify an etiology should include tests for a source of infection (including nasal swab, sputum/urine culture, and blood culture); tests for special pathogens (fungus, tuberculosis spot test); chest radiography (high-resolution CT); and bronchoscopy with BAL (17, 18, 20). Lung biopsy is not mandatory, and both drugs and infectious history can occasionally help to interpret results. Utilization of diagnostic tests is related to the suspected pneumonitis grade (6).

Common differential diagnoses for CIP include pulmonary infections, pulmonary embolism, DAD, lung cancer with underlying progression, cancerous lymphangitis, pulmonary interstitial edema caused by heart failure, fulminant myocarditis (120), and RIP (5, 23). Opportunistic pulmonary infections, including tuberculosis (TB) pneumonia, aspergillosis, cytomegalovirus pneumonia (CMVP), and Pneumocystis jirovecii pneumonia (PJP), have been the foremost differential diagnoses for CIP in the NSCLC population (26, 121–125). Inthasot et al. (26) reported two cases of severe lung infections complicating the treatment of nivolumab for NSCLC and emphasized the importance of eliminating the possibility of opportunistic infections. Notably, ICIs could cause special pathogen infections in some patients through induction of CIP. We included several cases of patients who developed CIP during ICI therapy and consequently developed rhinovirus/enterovirus (58), CMVP (62), PJP (62), legionella (25), human herpesvirus 6 (HHV-6) (73), pseudomonas, or candida (75) infections ( Figure 3 ). Because ICIs activate tumor immunity by inhibiting PD-1/PD-L1/CTLA-4, they might also simultaneously inhibit immunity to infection. Although the infections we described here as differential diagnoses are not usually categorized as drug-induced pneumonias, we included this series of reports to exemplify challenges in differentiating intensified infection from drug-induced pneumonia. Since from a drug safety perspective, the infection did lead to a few deaths.

The CT manifestations of CIP in patients with pulmonary AC sometimes resembled those of interstitial pneumonitis (126), especially of the OP pattern. Ichikawa et al. (127) reported that 2% of patients (13/564) with resected pulmonary AC presented with an OP pattern. Kanai et al. (79) reported a case of coexisting CIP and tumor invasion, which complicated the diagnosis and management of the lung disease. Aggressive lung biopsy was recommended in that study to correctly diagnose CIP in patients with NSCLC that mimicked the OP pattern or existed the tumor invasion.

RIP, an early lung injury induced by radiation, is also a difficult differential diagnosis in CIP. The approximate onset (1 to 3 months), similar imaging features (GGOs and diffuse haziness), and shared pathological feature (lymphocytic alveolitis) increased the level of challenge in distinguishing CIP from RIP (128–130). However, a distinct lesion location may assist in finding the difference between the two. RIP mainly exists in the radioactive region, and CIP mostly occurs outside the RT fall-off dose or in the low-dose field (48). Interestingly, both CIP and RIP have the same first-line therapy (corticosteroids) (121, 128). Meanwhile, radionics has emerged as a new approach to predict CIP by automatically extracting radiologic features for synthesis analysis (131).

In summary, CIP requires a precise diagnosis, including grade assessment, and monitoring of CIP requires a multidisciplinary method. Such monitoring often involves infectious disease specialists, pathologists, radiologists, pulmonologists, and cardiologists (121).

CIP is deemed a self-limiting disease. No prospective trials, to our knowledge, have evaluated the optimal therapeutic modality for CIP (5, 24). Current guidelines for CIP, therefore, recommended corticosteroids as the primary therapy approach (121, 132, 133). These decisions are based on the strength of case reports and clinical experience (5, 24). Different definitions of CIP grades are shown in Supplemental Table 2 (121, 133). Clinical improvement is usually observed after 48 to 72 hours of corticosteroid use, and patients without regression of CIP-related symptoms have been considered steroid refractory and treated with immunosuppressive agents (121, 133).

For patients with grade 1 CIP, clinical symptoms, imaging changes, and pulmonary function (diffusing capacity and spirometry) should be closely monitored for 3 to 4 weeks (122, 123, 134, 135). Tentatively stopping ICI treatment can be considered reasonable for mild cases of CIP (23). When the condition worsens, though, interruption of the ICI should be combined with initiation of low-dose steroids (0.5 to 1 mg/kg/d) (9, 136).

For patients with grade 2 CIP, withholding the ICIs and beginning intermediate-dose steroids (1 to 2 mg/kg/d) followed by a taper by 5 to 10 mg/week for 4 to 6 weeks have been proposed (133). In our analysis, we summarized the management characteristics stratifed by CIP grade ( Table 3 ) and listed every drug that every case used ( Figure 4 ). We converted the different steroid doses to methylprednisolone (MP) equivalents and divided these into three groups (low-dose, intermediate-dose, and high-dose groups) according to the initial equivalent administered at the beginning of the therapy. We also noticed that some cases did not describe the weight of patients, which led to two different specifications of steroid dose (mg/d and mg/kg/d). In patients with grade 2 CIP ( Table 3 ), 60% of patients were administered intermediate-dose steroids (60 to 500 mg/d). In other cases, 80% of patients with grade 2 CIP started with low-dose steroids (< 1 mg/kg/d). In addition, bronchoscopy and/or BAL plus initiation of empirical antibiotics when infection is suspected are recommended (14, 137). If clinical improvement does not happen after 2 to 7 days of monitoring, increasing the corticosteroid dose and adding immunosuppressive drugs should be considered (121, 138). Restarting ICI therapy may be considered when CIP is stable, has improved to grade ≤ 1, or has improved with 10 mg/d of prednisone (23). After re-initiation, physicians should evaluate clinical indicators every 3 days and perform chest imaging once a week to monitor for the flare and recurrence of CIP (9).

For patients with grade 3 to 4 CIP, ICI therapy should be discontinued immediately and permanently. The initial doses of steroids (1 to 2 mg/kg/d and 2 to 4 mg/kg/d) were approved and included in guidelines by the American Society of Clinical Oncology guidelines and the European Society for Medical Oncology (15, 121), respectively. However, no clinical trials have identified optimal corticosteroid doses or durations; therefore, therapy duration has always been adjusted largely on the basis of response to steroid treatment. Our analysis showed that patients with grade 3 or 4 CIP most often received glucocorticoid pulse therapy (44% of patients with grade 3 and 33% of patients with grade 4; Table 3 ). Initial steroid dosages of 1 to 2 mg/kg/d were mostly used in patients with severe CIP ( Table 3 ), and this dosage was consistent with the recommendations of the American Society of Clinical Oncology. Institutionally, we continue at the initial dosage until patients improve or remain stable (usually 1 week), at which time corticosteroids can be very slowly tapered during at least 5 to 8 weeks (9). Our data showed that the mean duration of a steroid taper was nearly 10 weeks, and the longest duration was in patients with grade 3 CIP (mean ± standard deviation of 16.37 ± 14.60 weeks) ( Table 3 ). Additional immunosuppressants, including infliximab (IFX), mycophenolate mofetil, intravenous immunoglobulin, tacrolimus, ciclosporin (57, 79), and cyclophosphamide, should be considered when the symptoms do not regress after 48 to 72 hours of treatment with corticosteroids (6, 14, 23, 137). Empirical antibiotics may be used to prevent opportunistic infection (122, 139–141). Our data also showed that the rates of immunosuppressive drug use (grade 2: 5.56%, grade 3: 21.05%, grade 4: 57.14%, P=0.016) and antibiotic use (grade 2: 22.22%, grade 3: 31.58%, grade 4: 85.71%, P=0.011) gradually increased with increasing severity of CIP ( Table 3 ).

Moreover, it has been reported that nearly one-fourth to one-third of patients experience CIP flares or recurrence after rapid corticosteroid tapers and appear recalcitrant to corticosteroid treatment (5). CIP recurrence may occur early in patients with more severe grade (grade 3 or 4) initially and have occurred most often in patients whose therapeutic course was shorter than 5 weeks (71, 142). The lengths of steroid courses from our data varied from 1 week to 73.4 weeks, and the mean duration for grade ≥ 2 CIP was more than 10 weeks ( Tables 2 and 3 ). However, in patients whose steroid course was shorter than 5 weeks (25, 49, 55, 59, 67, 72, 75, 82), two patients (25, 67) experienced CIP recurrence. The highest CIP recurrence rate, 22.22%, occurred in patients with grade 2 CIP ( Table 3 ). In addition, the steroid courses were centrally distributed in the first 5 weeks ( Figure 4 ), which suggests that the changes to steroid dosages (in grade 2 CIP) and drugs (in grade 3 or 4 CIP) usually occurred in this window.

Current experience with immunosuppressive drugs to treat CIP is based mostly on extrapolation from data about their use to treat other irAEs, which lacked pathophysiological evidence (5). IFX and cyclophosphamide have been approved to treat ICI-related digestive toxicities, especially colitis (133, 143, 144). However, IFX could itself cause ILD and liver injury (145–147). In addition, it could weaken the ongoing anticancer immune activity initially launched by ICI treatment (25); this hypothesis is consistent with a prior study (18), which reported that half of patients with grade 3 CIP died despite receiving additional immunosuppressive drugs. As a second-line drug, mycophenolate mofetil remains controversial because of its suppressive effects on the T-cell response (148). IL-17 blockade reportedly relieved ICI-related gastrointestinal and skin irAEs (149). Current guidelines also recommend cyclophosphamide, mycophenolate mofetil intravenously (1 g twice daily), or IFX (5 mg/kg) as supportive care (121, 133, 135) for steroid-resistant patients with irAEs. Intravenous immunoglobulin was effective in ICI-mediated myasthenia gravis and did not blunt infection responses (150). Thus, intravenous immunoglobulin could become a logical choice for treating CIP in patients with suspected comorbid infections (24). Tocilizumab, an IL-6 inhibitor, has been used to treat rheumatologic irAEs (84). A case report showed that a patient with NSCLC and CIP experienced significant symptom relief after additional therapy with tocilizumab (151). However, whether tocilizumab should be included as an option in the second-line drugs to treat steroid-refractory patients with irAEs remains undetermined, because that approach lacks a comparison with other second-line drugs.

Most studies have found that patients with CIP, especially with lower-grade disease, could see symptoms improve or resolve if they received corticosteroid therapies (18, 152). Similarly, our data ( Table 3 ) demonstrated that patients with grade 2 CIP all experienced improvements in or resolution of CIP and had the highest OS (85.71%) versus patients with grade 3 or 4 CIP (OS of 35.71% or 42.86%, respectively, P=0.019). In addition, nearly half of patients with grade 2 CIP experienced a partial tumor response, whereas most patients with grade 3 or 4 CIP experienced tumor progression or maintenance. However, a single-center study (20) recently reported poor prognoses in patients with NSCLC who developed CIP. Suresh et al. (45) demonstrated that the ICIs did not significantly influence the short-term survival (disease control rate, overall response rate, or PFS) but did affect OS which decreased by 10 months in patients with CIP. Fukihara et al. (47) came to a similar conclusion regarding the decrease in OS. Patients with CIP (8.7 months) had a shorter OS after PD-1 blockade compared with those without CIP (23.0 months, P=0.015). We also evaluated the association between CIP and OS ( Figure 5 ), and we found that patients who experienced deteriorated or maintained CIP were significantly more likely to have a poor prognosis compared with patients who experienced improved or resolved CIP (P=0.006). One potential reason might be that patients with CIP were more likely to be forced to quit ICI therapy to avoid lethal respiratory failure. Moreover, as a result of deteriorating physical status, abrasive pulmonary symptoms, and prolonged steroid management for CIP, patients with CIP tended to reject—and their physicians were more likely to hesitate or delay commencement or continuation of—aggressive anti-tumor treatment.

Recurrent phenomena related to the management of CIP have been explored in patients with NSCLC who received ICI therapy. These phenomena included recurrent pneumonia after completion of a steroid taper with or without restarting immunotherapy. Reports of reusing ICI therapy mainly occurred in patients with grade 1 or 2 CIP initially, since patients identified with grade 3 or 4 CIP generally withdrew treatment permanently (47). The reported recurrence ratio after reusing immunotherapy varied from 17% to 30% (18, 55, 122). Our analysis ( Table 2 ) showed that the overall recurrence ratios with and without re-challenge ICI therapy were 6.82% (3/44) and 22.73% (10/44), respectively. Among three patients (65, 73, 81) who re-challenged ICI therapy after clinical regression of CIP, two experienced recurrence after restarting ICIs (65, 81), and one patient successfully improved by discontinuing immunotherapy and beginning treatment with antibiotics and steroids (73). Recurrent pneumonitis severity, location of involvement, and pattern might vary compared with the initial manifestation of CIP.

Predictive factors for CIP are still under investigation. Currently, the exploration of serum markers, cytokines/chemokines, and cellular biomarkers have interested clinicians (137). Increased carcinoembryonic antigen (CEA) serum levels reportedly relate to both tumor progression and the simultaneous regression of recurrent CIP (153), which represents an early association with both durable toxicity and durable response. In addition, a low level of serum albumin was an independent predictor of CIP in patients with NSCLC (odds ratio=0.381, 95% CI=0.179–0.808, P=0.012) (71). In solid tumors, other research found that elevated baseline lymphocyte levels were linked to irAEs (47). In patients with melanoma who experienced severe irAEs, peripheral blood samples were evaluated early during treatment, and 11 elevated cytokines were recruited in the validation group for the predictive model (154).

We also evaluated the relationship between the initial steroid dose and OS ( Supplement Figures 1 , 2 ). Unfortunately, no significant difference in OS was found among low-dose, intermediate-dose, and high-dose steroid groups. Some reasons might be that the sample size was small and the precise data about steroid doses were limited, so the optimal steroid dose for OS was not determined. Therefore, extensive multicenter studies, which have detailed management of steroid therapy, should be conducted in the future.

The evolution of post-CIP patients is largely dependent on their CIP status. Patients with moderate or well-controlled CIP would have various subsequent treatment options including only supportive care, cytotoxic chemotherapy alone and ICIs rechallenge, based on the primary tumor response, irAEs evaluation, and patients’ willingness (155). Yamagata et al. (156) conducted a retrospective analysis concerning the NSCLC patients with CIP and reported the cancer therapy after CIP. They found that 34.6% of CIP patients decided to treat with cytotoxic chemotherapy, and 30.8% of CIP patients chose the best supportive care after CIP. The rechallenge of ICIs only applied on 3% of CIP patients. Actually, if the patients get complete or partial remission (CR or PR), the therapeutic strategies without ICIs could be considered for continued use (157). However, the options of rechallenge should be deliberated in the context of personalized consideration and multidisciplinary evaluation.

Patients with neurologic, cardiac, or any grade 4 irAEs are not recommended to continue or rechallenge ICIs (158). The evaluation of ICIs rechallenge mainly depends on risk-reward ratio (158). At present, there is no acknowledged guidance for re-challenging ICIs. Whether patients should resume ICI monotherapy after receiving doublet ICI therapy is still being investigated. A recent study recruited 80 patients with irAEs on doublet ICI therapy who subsequently reinstated ICIs as monotherapy, and the results indicated that the incidence of CIP (33%) was significantly higher than ophthalmic or gastrointestinal immune-related toxicity (159). However, in most instances, the ICI utilized for re-initiation in NSCLC could be the same ICIs used before, another PD-1/PD-L1 inhibitors, or the switching from PD-1 to PD-L1 inhibitors or the converse (160–164). Kitagawa et al. (157) included prior reports about ICIs rechallenge in NSCLC and analyzed its efficacy and safety. The results showed the generally lower overall response rate (ORR), disease control rate (DCR), and the median PFS presented in patients received the second ICI than in those received the first ICI among these studies. The greatest DCR (58.8%) and longest median PFS (4.0 months) during the second ICI treatment were showed in the 17 patients Kitagawa et al. (157) included. All these 17 patients switched the ICIs type when ICI rechallenge, of which 58.8% obtained PR or stable disease (SD) after switching ICIs administration. However, the efficacy of ICIs rechallenge is still controversial (165–167). Between two ICIs administration, shorter interval may exert better effects on outcome. Besides, the potential predictive factors of ICIs rechallenge outcome include early irAEs development, irAE therapy intensity, CIP phenotype, PD-1 inhibitors, and age more than 65 (155). As for the safety, the second attempt of ICIs could cause same irAEs or moderate new irAEs. Naidoo et al. (168) reported that 3 of 12 patients who reinstated ICI therapy developed CIP recurrence (initial CIP grade of 1 or 2), and 38 of 68 patients developed irAEs after re-treatment. Once patients experience recurrent CIP, the discontinued ICIs in time and the monotherapy of same steroid administrated before is universally acknowledged (19), while with a slower dose tapering and longer course (142).

Notably, there might exist durable anti-tumor activity after discontinuing ICIs therapy (44, 169, 170). This continuous treatment tendency could hold on until intolerable irAEs appearance, tumor progression or no more than 2 years. The correlation of tumor response and toxicity enhances the complexity of ICIs therapy and requires to be demonstrated further. Gauci et al. (170) found that the favorable predictive factors for prolonged response after stopping ICIs therapy included CR patients before discontinuation, with 13% increase of keeping disease stability compared to PR patients.

There are few reports about the sequela of CIP. The typical sequela might be the sustaining pulmonary interstitial fibrosis and poor pulmonary function caused by severe CIP (171, 172). Nintedanib, as an angiokinase blocker, has been reported to play significant role in progressive fibrosing interstitial lung disease, contributing to slow down the decline rate of forced vital capacity (FVC) (173) and further potentially strengthen the prevention of CIP (174).

Although quite a few researchers have intensively studied the characteristics of CIP in NSCLC, the studies with regard to the diagnosis, treatment and risk stratification require more exploration (175). First, timely and accurate diagnosis of CIP is necessary. The current biomarkers are based on the mechanism of irAEs. Among the various biomarkers, Isono et al. (176) recently found idiopathic interstitial pneumonias became the only risk factor of CIP in the multivariate Cox regression model. Therefore, the ability of these biomarkers to predict CIP should be investigated deeply.

Second, the management of CIP remains inconclusive. The optimal drug regimen of corticosteroid (taper and continuous time) for CIP and ICIs (onset) for post-CIP need more clinical studies with large sample size to evaluate. Currently, the corresponding two clinical protocols, NCT04036721 and NCT04169503, are ongoing and expected to present profound results.

Third, risk stratification for CIP contributes to precise treatment. CIP presents with different incidence and death rates in different histological types of NSCLC, which may be ascribed to the intrinsic features of tumor histological subtypes (19, 24). Thus, we need more research about the clinical, radiological, histological, and biological characteristics of CIP to determine whether specific subsets of patients should be treated prophylactically.

We conducted the descriptive analyses to delineate the baseline characteristics and the intergroup differences in different CIP outcomes and CIP grade groups. Kruskal-Wallis test and chi-square or Fisher’s exact test were utilized to analyze continuous and categorical variables, respectively. The former variables were presented by means and standard deviations, and the latter variables were expressed as counts and proportions. The overall and CIP survival rate were estimated by Kaplan-Meier method with a log-rank test. The statistical software packages R and EmpowerStats (X&Y Solutions Inc., Boston, MA, USA) were utilized to conduct all the statistical analyses. Two-sided P-values less than 0.05 were considered to identify statistical significance.

QZ and LT searched the literature and wrote the manuscript. YZ helped to collect literature and participated in discussions. LT and WH performed the statistics analysis. WL examined and verified the study. All authors contributed to the article and approved the submitted version.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.