Imaging features and prognostic significance of immune checkpoint inhibitor–related pneumonitis in NSCLC

Abstract

Background

Immune checkpoint inhibitors (ICIs) have demonstrated substantial therapeutic efficacy in the treatment of non-small cell lung cancer (NSCLC); however, their clinical application is associated with unique immune-related adverse effects (irAEs). Among these adverse events, immune checkpoint inhibitor-related pneumonitis (CIP) is rare yet serious, which may potentially result in severe respiratory failure, thereby requiring close clinical monitoring. Research specifically focusing on CIP in NSCLC patients treated with PD-1 inhibitors remain limited. This study targets this distinct cohort to comprehensively investigate the clinical and radiological determinants associated with overall survival, applying time-dependent covariate Cox regression to capture the dynamic impact of prognostic factors over time.

Methods

A total of 102 NSCLC participants who received immunotherapy with programmed cell death protein-1 (PD-1) inhibitors and then developed CIP were retrospectively enrolled in this study. Univariate and multivariate time-dependent covariate Cox regression models were constructed to determine associations between CIP features and survival benefits of CIP patients.

Results

The incidence of CIP was 15% (102/680) with a median onset time of 4.6 months. Fifty-one patients (50.0%) were identified as having organizing pneumonia (OP) pattern, followed by nonspecific interstitial pneumonia (NSIP) pattern in 28 patients (27.4%), hypersensitivity pneumonitis (HP) pattern in 6 patients (5.9%), and diffuse alveolar damage (DAD) pattern in 2 patients (2.0%). Additionally, 15 patients (14.7%) were classified as unclassifiable pattern. Kaplan-Meier analysis and Log-rank test indicated that CIP located around the tumor and with reticular opacity were associated with poorer prognosis (P = 0.023, P = 0.013). Compared to those with CIP grades 2–4, patients with CIP grade 1 demonstrated survival benefit with border-line significance (P = 0.049). Multivariate time-dependent covariate Cox regression analysis showed that CIP improvement or not (χ² = 6.81, P = 0.009), percentage of neutrophils (χ² = 24.13, P < 0.001) and albumin (χ² = 31.48, P < 0.001) at the time of CIP diagnosis were independent influencing factors for overall survival (OS) in NSCLC patients with CIP.

Conclusions

CIP without improvement or resolution, a high percentage of neutrophils and elevated albumin level of peripheral blood examination were independent predictors for the prognosis of NSCLC patients, which may have an implication for treatment.

Background

Immune checkpoint inhibitors (ICIs) have shown promising benefits and were extensively used in the treatment of various stages of non-small cell lung cancer (NSCLC), either in first-line or second-line setting [1, 2]. Despite emerging data that suggest beneficial results with ICIs in lung cancer treatment, it is worth noting that ICIs are associated with a unique spectrum of side effects termed immune-related adverse effects (irAEs), resulting from excessive immunity against normal organs [3]. These irAEs can occur in various organs, such as the skin, gastrointestinal system, endocrine system, neurological system, and lungs [4]. Among them, immune checkpoint inhibitor-related pneumonitis (CIP) is a relatively rare but serious irAE, potentially resulting in severe respiratory failure with possible discontinuation of therapy, or even life threatening which requires special attention [4,5,6,7].

Previous retrospective studies have reported considerable variation in the incidence and time of onset of CIP due to differences in the primary tumor site, the specific ICI agents evaluated (PD-1 inhibitors, PD-L1 inhibitors or CTLA-4 inhibitors), the type of regimen (monotherapy or multiagent regimen), and prior thoracic radiotherapy [8,9,10,11,12]. Compared to patients with other malignancies, higher prevalence and shorter onset time of CIP have been observed in patients with lung cancer treated with immunotherapy [13, 14]. Primary lung cancer patients may suffer from lung damage due to diverse factors, including smoking, pre-existing lung diseases, and the impact of radiation therapy [15,16,17]. This suggested a possible hypothesis that the characteristics of CIP in NSCLC patients may differ from those in other solid tumor patients [8]. Moreover, there existed highly controversial attitudes towards the impact of CIP on patients’ survival. In a cohort of 276 NSCLC patients who were treated with PD-1/PD-L1 inhibitors, CIP was associated with higher overall response rate and increased progression-free survival (PFS) [18], indicating an active immune status of those patients experiencing CIP. Nevertheless, other studies reported that the development of CIP was correlated with a poor prognosis in patients with NSCLC [19, 20].

The clinical and radiologic characteristics of CIP, along with their association with prognosis, have not been fully investigated in NSCLC patients [18, 21,22,23,24]. Therefore, we performed a retrospective study to explore clinical characteristics and CT imaging features of CIP, as well as its association with survival outcomes in NSCLC treated with PD-1 inhibitors.

Methods

Study design and participants

This study was approved by the institutional review board and conducted in accordance with the principles of the Declaration of Helsinki. Because of the retrospective nature of the study, informed consent was waived. Medical records of NSCLC patients treated with ICIs at Tianjin Medical University Cancer Institute & Hospital from January 2019 to December 2021 were retrieved from the institutional electronic medical database. Patients meeting the study criteria were consecutively enrolled. The inclusion criteria were as follows: (a) histologically or cytologically confirmed NSCLC without concurrent secondary primary tumor; (b) treated with PD-1 inhibitors as any line of therapy; (c) thin-section CT images of lung during ICI treatment were available. The exclusion criteria were as follows: (a) prior or combined thoracic radiotherapy including the lungs, mediastinum, thoracic spine, or ribs; (b) either an active autoimmune disease or a history of autoimmune disease which was in need of treatment; (c) no pneumonitis was suggested on follow up CT images; and (d) pneumonitis difficult to distinguish from CIP. Detailed inclusion and exclusion criteria are shown in Fig. 1.

[IMAGE OMITTED: SEE PDF]

All patients were regularly followed up from the time of initiation of ICI therapy until death or the last follow-up date (April 1 st, 2024). For patients who developed respiratory symptoms at any point during treatment, CT examination was performed to determine the presence of pneumonitis. Otherwise, regular clinical evaluations and CT scans after each two or three cycles of ICIs were conducted to evaluate treatment response. Overall survival (OS) was defined as the time from the first day of ICI initiation to the event of death from any cause or observational cessation.

Clinical information

All medical records of the patients before immunotherapy were retrieved, including age, gender, smoking history, drinking history, pathological type, PD-L1 expression, cancer stage and treatment strategy. Date of pneumonitis diagnosis, respiratory symptoms, and laboratory data of peripheral blood examination at the time of CIP diagnosis before steroids or other agents given to treat the CIP were recorded. Moreover, we computed the results of inflammatory biomarkers in peripheral blood, including neutrophil-to-lymphocyte ratio (NLR), platelet-to-lymphocyte ratio (PLR), lymphocyte-to-monocyte ratio (LMR), derived NLR (dNLR), and systemic immune-inflammation index (SII).

CT image acquisition

CT examinations were performed using one of the following CT scanners: GE Discovery CT750 HD, GE Optima 680 Expert, GE Revolution Evo, Simens SOMATOM Definition AS+, Simens SOMATOM Drive, and Philips IQon Spectral. Patients were in supine position and scanned from the entrance of the chest to the base of the diaphragm with deep inspiration breath-hold. The CT image acquisition parameters were as follows: tube voltage, 120 kVp; tube current, 150–200 mA or automatic tube current; reconstruction thickness, 1.25–1.5 mm; matrix, 512 mm × 512 mm; field of view (FOV), 36.5 cm to 44 cm. All images were displayed in standard lung (width, 1,500 HU; level, −500 HU) and mediastinal (width, 350 HU; level, 50 HU) window settings.

CT imaging assessment

According to the international guidelines released by the Fleischner Society, CIP was determined by the treating physician and the radiologists at the time of diagnosis based on clinical, radiological and necessary histological examination findings by excluding other potential known causes [5]. In the current study, CIP is defined as follows: (a) appearance of new pulmonary parenchymal opacities at imaging after initiation of ICIs and (b) identification that it is not an infection, tumor progression, diffuse alveolar hemorrhage, or heart failure by the investigator. The diagnosis of CIP was based on assessments by radiologists, aetiological examination, peripheral blood examination, clinical records and necessary histological examination or multidisciplinary discussion.

The chest CT images were independently reviewed by two radiologists (X.Q.W., Y.L., with experience of 3 and 19 years in chest imaging, respectively) both of whom were blinded to the clinical data and outcomes of the patients. Any discrepancy was resolved by discussion to reach a consensus.

Based on the American Society of Clinical Oncology (ASCO) Guideline grading system which include clinical and imaging manifestations [6], CIP was classified into four grades: grade 1 (asymptomatic; confined to one lung lobe or < 25% of lung parenchyma), grade 2 (symptomatic; involves more than one lung lobe or 25-50% of lung parenchyma), grade 3 (severe symptoms; involves all lung lobes or > 50% of lung parenchyma; oxygen indicated), and grade 4 (life-threatening respiratory compromise).

“CIP surrounding tumor” is defined as follows: (a) CIP immediately adjacent to the primary lung cancer and the boundary is not clear and (b) with or without CIP located distant from the tumor. “CIP distant from the tumor” is defined as follows: (a) no direct contact between CIP and primary lung cancer; (b) patients who underwent lung surgery before immunotherapy.

CIP imaging patterns

The radiologic patterns of CIP were divided into five subtypes [5, 25]: organizing pneumonia (OP), nonspecific interstitial pneumonia (NSIP), hypersensitivity pneumonitis (HP), diffuse alveolar damage (DAD), and unclassifiable.

Organizing pneumonia (OP) pattern is characterized by multifocal patchy alveolar opacities, typically distributed peribronchovascular or peripheral, and it may display reversed halo sign (Fig. 2a, Supplementary Fig. 1). Additional suggestive features include band-like or perilobular consolidations [5, 25]. Occasionally, it might present as solitary or multiple nodular opacities [26].

[IMAGE OMITTED: SEE PDF]

Nonspecific interstitial pneumonia (NSIP) pattern consists of patchy areas of ground-glass opacity (GGO), usually more prominent in the periphery and lower lung zone. It may manifest progression to reticular opacities and traction bronchiectasis, sometimes accompanied by areas of consolidation. In some patients, subpleural sparing of the posterior has also been observed (Fig. 2b, Supplementary Fig. 2) [5, 25].

Hypersensitivity pneumonitis (HP) pattern shows poorly defined small centrilobular nodules, sometimes accompanied by extensive areas of GGO or decreased attenuation and vascularity (mosaic attenuation) (Fig. 2c, Supplementary Fig. 3). It is usually more prominent in the upper lobe [5, 25].

Diffuse alveolar damage (DAD) pattern shows widespread bilateral areas of GGO and consolidation in exudative phase, along with traction bronchiectasis and reduced lung volumes in fibrotic phases (Fig. 2d, Supplementary Fig. 4) [5, 25].

Unclassifiable pattern included bronchiolitis and abnormalities lacking typical aspect or distribution features, such as irregular or strip-like GGO (Fig. 2e, f; Supplementary Fig. 5, 6) [5, 25]. Additionally, cases exhibiting a concurrent presence of two distinct radiological pattern were also categorized under this pattern.

Statistical analysis

SPSS software (Version 26.0) and R language (Version 4.4.0) were used for statistical analysis. Continuous variables were expressed as mean ± standard deviation or median (interquartile range, IQR) and categorical variables were expressed as a number (percentage). To assess the inter-reader agreement for radiological pattern classification, Cohen’s Kappa statistics were calculated (k = 0.00–0.20, poor; 0.21–0.40, fair; 0.41–0.60, moderate; 0.61–0.80, substantial; and 0.81–1.00, almost perfect). Kaplan-Meier analysis and Log-rank test were applied to assess survival outcomes. The Cox regression model is utilized only when the hazard ratio (HR) remains constant over time. The proportional hazard hypothesis for different variables can be evaluated using time-dependent covariate methods. P < 0.05 for the interaction term for time indicates a violation of the proportional hazard hypothesis, thereby necessitating the use of time-dependent covariate Cox regression model. Variables with P < 0.05 in univariate analysis were included in multivariate Cox regression analysis using backward stepwise method. Time‐dependent covariate Cox regression model was applied to estimate HR and identify prognostic factors for OS. All P-values were two-sided hypothesis, and P < 0.05 was considered statistically significant.

Handling of missing data

Given the inevitable problem of missing data, we used single regression-based imputation to resolve the issue of missing value filling (six patients had missing values for peripheral blood cell examinations, and seven patients had missing values for liver function test). This method was selected to address small proportions of randomly missing continuous variables.

Results

Clinical characteristics

A total of 680 NSCLC patients treated with PD-1 inhibitors without thoracic radiotherapy were screened out according to the criteria, and among them, 102 patients (15%) developed CIP. The demographic, clinical, and pathological characteristic details of CIP patients are shown in Table 1. Of these patients with CIP, 85 (83.3%) were males and 17 (16.7%) were females. The median age was 63 years (57–68 years). Eighty-six cases (84.3%) were former or current smokers, and drinking history was detected in 53 cases (52.0%). The majority of tumors were adenocarcinomas (n = 51, 50.0%), followed by squamous cell carcinoma (n = 41, 40.2%). There were 6 patients (5.9%) with early stage (I-II) NSCLC and 96 patients (94.1%) with advanced stage (III-IV). All patients were treated with PD-1 inhibitors either monotherapy (n = 14, 13.7%) or combination therapy (n = 88, 86.3%). The median medication duration of immunotherapy was 8.9 months (4.1–17.6 months). In addition, a total of 29 patients experienced other irAEs, including rash, hypothyroidism, arthritis, and others. Detailed information about these patients can be found in Supplementary Table 1.

[IMAGE OMITTED: SEE PDF]

The median onset time from the initiation of ICI therapy to CIP is 4.6 months (1.6–7.7 months) and the median medication cycle is 6 (2–10 cycles). Fifty patients (49.0%) were asymptomatic at the onset of CIP, while the remaining 52 (51.0%) patients had clinical symptoms. The most common symptoms included cough (36 patients, 35.3%), dyspnea (23 patients, 22.5%), expectoration (23 patients, 22.5%) and other patients might suffer from fever, hemoptysis, or shortness. Some cases had multiple clinical manifestations.

Radiologic characteristics

All CIP patients had chest CT evaluations at the prior and onset of CIP. The spectrum of radiologic manifestations of CIP are shown in Table 2. Specifically, GGO was the most common feature (98/102, 96.1%), followed by consolidation (55/102, 53.9%), interlobular septal thickening (49/102, 48.0%), and reticular opacity (41/102, 40.2%); other relatively rare radiologic findings included nodular opacity, pleural effusion, bronchitis, and bronchiectasis. Almost perfect interobserver agreement was achieved between the two readers regarding the classification of CIP radiological pattern (k = 0.89). OP (51 patients, 50.0%) was the predominant radiologic pattern, followed by NSIP (28 patients, 27.4%) as the second most common pattern; while HP (6 patients, 5.9%) and DAD (2 patients, 2.0%) patterns were relatively rare. In addition, 15 patients (14.7%) were classified as unclassifiable. For subgroup analysis, OP pattern was still the most prevalent radiologic pattern among different CIP grades (Supplementary Table 2) and no significant difference was observed in CIP patterns between different CIP grades (P = 0.192).

[IMAGE OMITTED: SEE PDF]

Of the 102 patients with CIP, 25 of them underwent surgical excision of the lung tumor before immunotherapy (2 patients with one lobe resection and wedge resection of another lobe, 17 patients with one lobe resection, 5 patients with wedge resection of one lobe, and 1 patient with wedge resection of two lobes), and the CIP distribution was classified as “distant from tumor” for these patients. For those with primary lung lesions, nearly half of them showed CIP around the lung tumor (38/77, 49.4%). CIP occupying the same side of the primary tumor was found in 19 patients (19/102, 18.6%), the opposite side in 16 patients (16/102, 15.7%) and involving both sides of lung fields in 67 patients (67/102, 65.7%).

Association of CIP with survival benefit

From the occurrence of CIP to the last follow-up time (April 1 st, 2024), 57 out of 102 CIP patients died, with a mortality rate of 55.9%. The median follow-up time from baseline to last follow-up was 18.6 months (8.6–33.4 months), and OS was 29.4 ± 16.1 months for the entire cohort. Compared to those with CIP grades 2–4, patients with CIP grade 1 demonstrated survival benefit with border-line significance (P = 0.049). However, no significant difference was observed between the grades 1–2 and grades 3–4 groups (P = 0.119) (Fig. 3). Compared to patients with CIP surrounding tumor and reticular opacity at the time of CIP diagnosis, those with CIP distant from tumor and without reticular opacity demonstrated survival benefit (P = 0.023, P = 0.013) (Fig. 4). For those 77 patients with primary lung lesions, there was no significant difference in overall survival between CIP distant from tumor and CIP surrounding the tumor (P = 0.066) (Supplementary Fig. 7).

[IMAGE OMITTED: SEE PDF]

[IMAGE OMITTED: SEE PDF]

Treatment and resolution

One patient died six days after CIP onset due to severe hemoptysis without CT imaging follow-up, while for the remaining 101 patients, the median imaging follow-up time was 12.7 months (5.5–23.5 months). According to the Fleischner Society Position Paper [5], patients with severe or progressive lung disease should be advised to discontinue the suspected drug. Twenty patients with CIP continued to receive ICI therapy without additional therapy, and close monitoring with physical examination and chest CT was recommended. Among these patients, imaging follow-up revealed CIP improvement or resolution in 12 patients, persistent in 1 patient, migratory in 2 patients, and worsened in the remaining 5 patients. Eighty-one patients were temporarily or permanently discontinued immunotherapy, and whether corticosteroids were needed depends on the treating physician. Of those 81 patients, 50 patients had radiographic improvement or resolution (Supplementary Fig. 1–3) during the follow-up course, 4 patients showed persistent CIP, 15 patients had CIP with migratory (Supplementary Fig. 4), and 12 patients developed worsening and progression of CIP with new pneumonia infiltration lesions on CT images (Supplementary Fig. 8).

Univariate time-dependent covariate Cox regression

Univariate time-dependent covariate Cox regression analysis was conducted to examine whether the independent variables met the proportional hazard hypothesis and to explore the factors influencing OS in CIP patients. Time-dependent covariates were constructed to assess the proportional hazard hypothesis. The impact of CIP outcome, percentage of neutrophils, percentage of basophils, and albumin on survival time varies over time. The above indicators did not meet the proportional hazard hypothesis (P < 0.05), suggesting that these variable were time-dependent covariables. All other factors were included in the univariate Cox proportional hazards regression analysis. Univariate time-dependent covariate Cox regression analysis indicated CIP improvement or resolution, percentage of neutrophils, percentage of basophils, and albumin were prognostic factors (P < 0.05) (Supplementary Table 3, Table 3). Univariate Cox proportional hazards regression analysis revealed that CIP onset months, CIP location, CIP with reticular opacity or not, and globulin showed significant associations with overall survival (P < 0.05) (Supplementary Table 4, Table 4).

[IMAGE OMITTED: SEE PDF]

[IMAGE OMITTED: SEE PDF]

Multivariate time-dependent covariate Cox regression

Multivariate time-dependent covariate Cox regression analyses revealed that CIP improvement or resolution (χ2 = 6.81, P = 0.009), low percentage of neutrophils (χ2 = 24.13, P < 0.001) and albumin (χ2 = 31.48, P < 0.001) at the time of CIP diagnosis were independent predictors of a longer OS in NSCLC patients with CIP (Table 5). The CIP improvement or resolution×ln(OS) (B=−2.31), percentage of neutrophils×ln(OS) (B=−0.25), and albumin×ln(OS) (B=−0.64) had negative B values, indicating that the risk of death associated with these factors decreases over time in NSCLC patients with CIP. For CIP not improving or resolution, HR = Exp(7.01–2.31×ln(OS)), while for one-unit increase in the percentage of neutrophils, HR = Exp(0.83 − 0.25×ln(OS)), and for one-unit increase in albumin, HR = Exp(2.00–0.64×ln(OS)). All of these are time-dependent functions.

[IMAGE OMITTED: SEE PDF]

Discussion

In this retrospective study, we revealed CT imaging features of CIP and investigated the relationship between CIP features and survival benefits in NSCLC patients treated with PD-1 inhibitors in a clinical setting. A higher incidence of CIP (15%) than those reported in most of previous studies (0.16 to 11.8% of all malignancies) [13, 27, 28] was observed in the current cohort. This can be attributed to the following reasons: (1) It has been proposed that the incidence of CIP differs depending on the type of ICI, with a higher frequency observed in patients receiving PD-1 inhibitors compared to those treated with CTLA-4 inhibitors or PD-L1 inhibitors [9, 10]. (2) The majority of patients (84.3%) were current or former smokers in our study, giving rise to underlying preexisting pulmonary diseases [15, 29]. (3) We restricted the enrolled patients with NSLCL, of whom might have a lower pretreatment pulmonary reserve [14, 30]. (4) In our study, most patients (68/102, 66.7%) were first line treatment, and it has been found that this status was associated with a higher risk of CIP, suggesting that clinical course may play a role in the development of CIP [9].

The clinical manifestations of CIP were often nonspecific. The most common clinical manifestations were cough, followed by dyspnea or expectoration. Nearly half of the patients were asymptomatic at the onset of CIP, which was higher compared to other studies [13, 21]. Regular imaging follow-up was conducted at our institution to evaluate treatment response, thereby increasing the opportunity to identify a part of asymptomatic low-grade CIP cases.

In our study, the median onset time of CIP was 4.6 months, which was longer than previous studies (median onset time within 3 months) [8, 13]. This may be attributed to the extended duration of imaging follow-up (median: 12.7 months) which was helpful to identify late-onset CIP. Huang et al. have showed that the overall survival rate of the early-onset group (CIP occurring within 6 weeks of ICI treatment) was significantly lower than that of the late-onset group (CIP occurring after 6 weeks of ICI treatment) (P < 0.05) [22]. Similarly, we noticed that the risk of death decreases by 15% with each unit increase in the CIP onset time in univariate Cox proportional hazards regression analyses (HR = 0.85, 95% CI: 0.79–0.93, P < 0.001), although it did not become an independent risk factor for OS in multivariate Cox regression analysis.

Our results showed that GGO was the most common radiologic manifestation of CIP, followed by consolidation, interlobular septal thickening and reticular opacity. OP pattern was most commonly seen among five subtypes of radiologic patterns of CIP, which was consistent with some of the published data [8, 31]. In particular, CIP involved both lungs and demonstrated diffuse distribution in most of the patients, which was in line with previous studies as well [13, 21, 22].

The grading of CIP was defined according to the ASCO Guideline grading system in the current study, which includes both clinical and imaging manifestations [6]. Kaplan-Meier curves indicated that compared to patients with CIP grades 2–4, patients with CIP grade 1 demonstrated survival benefit (P = 0.049), while in the univariate Cox regression analyses, the differences between CIP grade did not show statistical difference (P = 0.055). This may be attributed to the differences in the nature and assumptions of the two methods. The Kaplan-Meier method is capable of detecting survival differences between groups even with a small sample size. In contrast, insufficient events may lead to reduced power of the Cox regression model [32]. The results of this study revealed that there was no significant difference between grades 1–2 and 3–4 groups (P = 0.119). Cui et al. obtained slightly different results from ours that patients with grades 1–2 CIP had an increased PFS compared to patients without CIP (P < 0.01), whereas grades 3–4 CIP was not significantly associated with increased PFS [18]. Noteworthily, the divergence could partially originate from the difference of criteria used to grade CIP, since the National Cancer Institute’s Common Terminology Criteria for Adverse Events (CTCAE) version 4.03.10 that they used was solely based on clinical status. Also, the endpoint of their study was PFS, which was less convincing than OS.

Under the motivation that CT imaging characteristics such as CIP location, radiologic manifestation and dynamic change or clinical outcome of CIP may serve as prognostic indicators, we conducted innovative explorations in this regard. Kaplan-Meier curves indicated that patients with CIP located around the tumor had significantly shorter OS compared to patients with CIP located distant from the tumor (P = 0.023). Patients with lung tumor resection before immunotherapy that had decreased tumor burden were all located to “CIP distant from tumor” group might be one possible explanation for this phenomenon. Meanwhile, the radiologic manifestation of CIP without reticular opacity indicated better survival outcome (P = 0.013). Unfortunately, they did not become independent risk factors for OS in multivariate Cox regression analysis. Multivariate time-dependent covariate Cox regression analyses revealed that CIP improvement or resolution (χ2 = 6.81, HR = Exp(7.01–2.31×ln(OS)), P = 0.009) was an independent predictor of longer OS. Over time, the resolution of CIP is associated with a declining trend in mortality risk among NSCLC patients. This reduction may be attributed to therapeutic interventions typically administered by clinicians for patients with persistent or progressive CIP. To prevent fatal respiratory failure resulting from CIP progression, some of these patients required permanent discontinuation of ICI therapy, which may adversely impact long-term outcomes.

Up till now, only a few studies have been conducted to explore the potential predictive value of peripheral blood biomarkers in the prognosis of NSCLC patients with CIP. Since peripheral blood examination is a minimally invasive and commonly ordered laboratory test, we attempted to identify the potential risk factors in peripheral-blood at the time of CIP diagnosis which might correlated with the survival in CIP patients. Multivariate time-dependent covariate Cox regression analyses revealed that low albumin (χ2 = 31.48, P < 0.001) and percentage of neutrophils (χ2 = 24.13, P < 0.001) at the time of CIP diagnosis were independent predictors of a longer OS. The risk of death associated with these factors decreases over time in NSCLC patients with CIP. Hypoalbuminemia is associated with inflammation [33], which may suggest that decreased serum albumin is associated with the occurrence of irAEs. It remains unclear whether the decrease in serum albumin is a consequence of the inflammatory complications induced by ICIs or if the reduced serum albumin levels contribute to the development of CIP. Previous research showed that irAEs generally represent an active immune state and indicate an improved response to ICI treatment [34, 35]. The decrease in serum albumin at the time of CIP diagnosis may be related to the severity of immune-related inflammation. Conversely, Lin et al. have found that high albumin (≥ 33.8 g/L) (HR = 0.28, 95% CI: 0.08–0.94; P = 0.04) at the time of CIP diagnosis was associated with longer OS in advanced NSCLC patients with CIP [36]. However, their study did not test the proportional hazard hypothesis, while in our study, the risk associated with albumin changes over time. Jane et al. have demonstrated that circulating neutrophils (HR = 1.08, 95% CI: 1.05–1.11; P < 0.001) were significantly and independently associated with poorer prognosis and proposed that neutrophils play a role in promoting immune mediated tumor progression as well as cancer cachexia syndrome [37]. To better elucidate the mechanisms underlying peripheral blood biomarkers variation and its role in CIP, prospective studies with a larger sample size are needed.

Our study had some limitations. First, it was a retrospective study, and patients were diagnosed with CIP by excluding other potential known causes, only a small number had histological examination. According to current clinical guidelines, CIP is considered a diagnosis of exclusion, and histological confirmation is often not pursued due to the invasive nature of lung biopsy, potential risk to patient safety, and the limited diagnostic yield, particularly in patients with poor general condition. Second, although the chest CT images were independently reviewed by two radiologists and demonstrated substantial inter-observer agreement, a degree of subjectivity in imaging interpretation remains inevitable. Third, potential drug-related pneumonitis caused by other medications in combination therapy or other treatments administered before or after CIP onset, including antibiotics and supportive therapies, cannot be completely ruled out and may have introduced confounding factors. Fourth, because of the low incidence of CIP, the sample size was relatively small but still higher than most similar studies. Furthermore, the lack of unified treatment methods for CIP may lead to various medication duration, which could impact patient survival. Further multi-center studies with larger numbers of NSCLC patients are needed to validate our results.

Conclusions

In summary, our findings showed that OP pattern was the dominant radiographic pattern of CIP. When CIP develops in clinical settings, close attention should be given to the peripheral blood examination and pneumonitis outcome, because CIP improvement or resolution, low percentage of neutrophils and albumin at the time of CIP diagnosis could be served as signs of increased survival in NSCLC patients treated with PD-1 inhibitors. Based on these findings, we recommend that: (1) Patients with CIP should undergo regular imaging surveillance to monitor the evolution of pneumonitis; (2) Laboratory parameters such as neutrophil percentage and albumin levels should be closely monitored during CIP management; (3) Prompt clinical interventions should be implemented upon detection of abnormalities to improve CIP outcomes.

Data availability

The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Abbreviations

ICI:

Immune checkpoint inhibitor

NSCLC:

Non-small cell lung cancer

CIP:

Immune checkpoint inhibitor-related pneumonitis

PD-1:

Programmed cell death protein-1

OP:

Organizing pneumonia

NSIP:

Nonspecific interstitial pneumonia

HP:

Hypersensitivity pneumonitis

DAD:

Diffuse alveolar damage

GGO:

Ground-glass opacity

IQR:

Interquartile range

OS:

Overall survival

HR:

Hazard ratio

CI:

Confidence intervals

References

Remon J, Soria JC, Peters S. Early and locally advanced non-small-cell lung cancer: an update of the ESMO clinical practice guidelines focusing on diagnosis, staging, systemic and local therapy. Ann Oncol. 2021;32:1637–42. https://doi.org/10.1016/j.annonc.2021.08.1994.

Miller M, Hanna N. Advances in systemic therapy for non-small cell lung cancer. BMJ. 2021;375:n2363. https://doi.org/10.1136/bmj.n2363.

Postow MA, Sidlow R, Hellmann MD. Immune-related adverse events associated with immune checkpoint blockade. N Engl J Med. 2018;378:158–68. https://doi.org/10.1056/NEJMra1703481.

Ramos-Casals M, Brahmer JR, Callahan MK, et al. Immune-related adverse events of checkpoint inhibitors. Nat Rev Dis Primers. 2020;6:38. https://doi.org/10.1038/s41572-020-0160-6.

Johkoh T, Lee KS, Nishino M, et al. Chest CT diagnosis and clinical management of Drug-Related pneumonitis in patients receiving molecular targeting agents and immune checkpoint inhibitors: A position paper from the Fleischner society. Chest. 2021;159:1107–25. https://doi.org/10.1016/j.chest.2020.11.027.

Schneider BJ, Naidoo J, Santomasso BD, et al. Management of immune-related adverse events in patients treated with immune checkpoint inhibitor therapy: ASCO guideline update. J Clin Oncol. 2021;39:4073–126. https://doi.org/10.1200/JCO.21.01440.

Haanen J, Obeid M, Spain L, et al. Management of toxicities from immunotherapy: ESMO clinical practice guideline for diagnosis, treatment and follow-up. Ann Oncol. 2022;33:1217–38. https://doi.org/10.1016/j.annonc.2022.10.001.

Nobashi TW, Nishimoto Y, Kawata Y, et al. Clinical and radiological features of immune checkpoint inhibitor-related pneumonitis in lung cancer and non-lung cancers. Br J Radiol. 2020;93: 20200409. https://doi.org/10.1259/bjr.20200409.

Khunger M, Rakshit S, Pasupuleti V, et al. Incidence of pneumonitis with use of programmed death 1 and programmed death-Ligand 1 inhibitors in Non-Small cell lung cancer: A systematic review and Meta-Analysis of trials. Chest. 2017;152:271–81. https://doi.org/10.1016/j.chest.2017.04.177.

Berz AM, Boughdad S, Vietti-Violi N, et al. Imaging assessment of toxicity related to immune checkpoint inhibitors. Front Immunol. 2023;14:1133207. https://doi.org/10.3389/fimmu.2023.1133207.

Chen X, Zhang Z, Hou X, et al. Immune-related pneumonitis associated with immune checkpoint inhibitors in lung cancer: a network meta-analysis. J Immunother Cancer. 2020; 8. https://doi.org/10.1136/jitc-2020-001170.

Tan P, Huang W, Wang L, et al. Deep learning predicts immune checkpoint inhibitor-related pneumonitis from pretreatment computed tomography images. Front Physiol. 2022;13:978222. https://doi.org/10.3389/fphys.2022.978222.

Nishino M, Ramaiya NH, Awad MM, et al. PD-1 Inhibitor-Related pneumonitis in advanced Cancer patients: radiographic patterns and clinical course. Clin Cancer Research: Official J Am Association Cancer Res. 2016;22:6051–60.

Ma K, Lu Y, Jiang S, Tang J, Li X, Zhang Y. The relative risk and incidence of immune checkpoint inhibitors related pneumonitis in patients with advanced cancer: a meta-analysis. Front Pharmacol. 2018;9:1430. https://doi.org/10.3389/fphar.2018.01430.

Ma Z, Lv J, Zhu M, et al. Lung cancer risk score for ever and never smokers in China. Cancer Commun (Lond). 2023;43:877–95. https://doi.org/10.1002/cac2.12463.

Brenner DR, Boffetta P, Duell EJ, et al. Previous lung diseases and lung cancer risk: a pooled analysis from the international lung Cancer consortium. Am J Epidemiol. 2012;176:573–85.

Huang Y-J, Huang T-W, Lin F-H, Chung C-H, Tsao C-H, Chien W-C. Radiation therapy for invasive breast cancer increases the risk of second primary lung cancer: a nationwide population-based cohort analysis. J Thorac Oncol. 2017;12:782–90. https://doi.org/10.1016/j.jtho.2017.01.021.

Cui P, Huang D, Wu Z, et al. Association of immune-related pneumonitis with the efficacy of PD-1/PD-L1 inhibitors in non-small cell lung cancer. Ther Adv Med Oncol. 2020;12:1758835920922033. https://doi.org/10.1177/1758835920922033.

Suresh K, Psoter KJ, Voong KR, et al. Impact of checkpoint inhibitor pneumonitis on survival in NSCLC patients receiving immune checkpoint immunotherapy. J Thorac Oncol. 2019;14:494–502. https://doi.org/10.1016/j.jtho.2018.11.016.

Tiu BC, Zubiri L, Iheke J, et al. Real-world incidence and impact of pneumonitis in patients with lung cancer treated with immune checkpoint inhibitors: a multi-institutional cohort study. J Immunother Cancer. 2022. https://doi.org/10.1136/jitc-2022-004670.

Pang L, Xie M, Ma X, et al. Clinical characteristics and therapeutic effects of checkpoint inhibitor-related pneumonitis in patients with non-small cell lung cancer. BMC Cancer. 2023;23:203. https://doi.org/10.1186/s12885-023-10649-0.

Huang A, Xu Y, Zang X, et al. Radiographic features and prognosis of early- and late-onset non-small cell lung cancer immune checkpoint inhibitor-related pneumonitis. BMC Cancer. 2021;21:634. https://doi.org/10.1186/s12885-021-08353-y.

Watanabe S, Ota T, Hayashi M, et al. Prognostic significance of the radiologic features of pneumonitis induced by anti-PD-1 therapy. Cancer Med. 2020;9:3070–7. https://doi.org/10.1002/cam4.2974.

Chu X, Zhao J, Zhou J, et al. Association of baseline peripheral-blood eosinophil count with immune checkpoint inhibitor-related pneumonitis and clinical outcomes in patients with non-small cell lung cancer receiving immune checkpoint inhibitors. Lung Cancer. 2020;150:76–82. https://doi.org/10.1016/j.lungcan.2020.08.015.

Pozzessere C, Lazor R, Jumeau R, Peters S, Prior JO, Beigelman-Aubry C. Imaging features of pulmonary immune-related adverse events. J Thorac Oncol. 2021;16:1449–60. https://doi.org/10.1016/j.jtho.2021.05.017.

Larsen BT, Chae JM, Dixit AS, Hartman TE, Peikert T, Roden AC. Clinical and histopathologic features of immune checkpoint inhibitor-related pneumonitis. Am J Surg Pathol. 2019;43:1331–40. https://doi.org/10.1097/PAS.0000000000001298.

Zhou P, Zhao X, Wang G. Risk factors for immune checkpoint inhibitor-related pneumonitis in cancer patients: a systemic review and meta-analysis. Respiration. 2022;101:1035–50. https://doi.org/10.1159/000526141.

Naidoo J, Wang X, Woo KM, et al. Pneumonitis in patients treated with anti-programmed death-1/programmed death ligand 1 therapy. J Clin Oncol. 2017;35:709–17. https://doi.org/10.1200/JCO.2016.68.2005.

Chen X, Sheikh K, Nakajima E, et al. Radiation versus immune checkpoint inhibitor associated pneumonitis: distinct radiologic morphologies. Oncologist. 2021;26: e1822–32. https://doi.org/10.1002/onco.13900.

Gomatou G, Tzilas V, Kotteas E, Syrigos K, Bouros D. Immune checkpoint inhibitor-related pneumonitis. Respiration. 2020;99:932–42. https://doi.org/10.1159/000509941.

Delaunay M, Cadranel J, Lusque A, et al. Immune-checkpoint inhibitors associated with interstitial lung disease in cancer patients. Eur Respir J. 2017. https://doi.org/10.1183/13993003.00050-2017.

Frank EH. Regression modeling strategies with applications to linear models, logistic and ordinal regression, and survival analysis, Spinger, 2015.

Soeters PB, Wolfe RR, Shenkin A. Hypoalbuminemia: pathogenesis and clinical significance. JPEN J Parenter Enter Nutr. 2019;43:181–93. https://doi.org/10.1002/jpen.1451.

Fujii T, Colen RR, Bilen MA, et al. Incidence of immune-related adverse events and its association with treatment outcomes: the MD Anderson Cancer center experience. Investig New Drugs. 2018;36:638–46. https://doi.org/10.1007/s10637-017-0534-0.

Sugano T, Seike M, Saito Y, et al. Immune checkpoint inhibitor-associated interstitial lung diseases correlate with better prognosis in patients with advanced non-small-cell lung cancer. Thorac Cancer. 2020;11:1052–60. https://doi.org/10.1111/1759-7714.13364.

Lin X, Deng H, Yang Y, et al. Peripheral blood biomarkers for early diagnosis, severity, and prognosis of checkpoint inhibitor-related pneumonitis in patients with lung cancer. Front Oncol. 2021;11: 698832. https://doi.org/10.3389/fonc.2021.698832.

Margetts J, Ogle LF, Chan SL, et al. Neutrophils: driving progression and poor prognosis in hepatocellular carcinoma? Br J Cancer. 2018;118:248–57. https://doi.org/10.1038/bjc.2017.386.