Introduction

Anti-programmed cell death 1 (PD-1)/anti-PD ligand 1 (PD-L1) immune checkpoint inhibitors (ICIs) improve the prognosis of patients with various malignancies. However, the occurrence of immune-related adverse events (irAEs) cannot be ignored. Although most irAEs are managable, some are fatal, e.g., checkpoint inhibitor pneumonitis and myocarditis [1,2,3,4]. The overactivation of T cells that recognize self-proteins or commensal microorganisms is presumed to be the probable pathogenesis of irAEs [5, 6]. Cluster of differentiation 8-positive (CD8+) lymphocyte infiltration in irAE-targeted tissues and/or organs was shown in patients with checkpoint inhibitor-induced colitis and myocarditis [6,7,8]. However, the features of T-cell subsets associated with checkpoint inhibitor pneumonitis (CIP) were not consistent across different studies. Bukhari et al. analyzed peripheral blood T-cell subtypes and reported that different populations of T cells were associated with pulmonary radiographical-immunological correlation [5]. Suresh et al. reported that bronchoalveolar lavage fluid (BALF) analysis of CIP patients revealed lymphocytosis and predominantly CD4+ T cells [9]. Compared with the control group (cancer patients who had received immunotherapy and did not develop CIP), BALF CD8+ T cell percentage of live cells was also higher in the CIP group, but failed to show statistical significance. And the median percentages of CD4+ T cells and CD8+ T cells were close (both were around 10% of live cells). However, the proportion of CD8+ cells among T cells was significantly or tended to be higher for CIP than for the other types of ILD in Suzuki’s CIP study [10].

Lung biopsies are not commonly performed for CIP patients [11], and BALF analysis is a useful diagnostic procedure for the differentiation of interstitial lung diseases (ILDs), infectious pulmonary diseases or pulmonary malignancies, especially for fragile patients and/or patients at high risk for severe complications after percutaneous or surgical lung biopsy [12, 13]. Elucidation of the characteristics of inflammatory cells in BALF might help us to understand the underlying pathophysiological process of CIPs. To date, the exact etiology and pathogenesis of CIP have not been determined. However, detailed studies on the BALF characteristics of CIP patients have rarely been reported. Therefore, we conducted this retrospective study to describe the inflammatory cellular patterns and T lymphocyte subset features in the BALF of CIP patients.

Patients and methods

Patients

There were 154 patients with pathology-confirmed malignancies suffering from CIPs after ICI therapy who were admitted to Peking Union Medical College Hospital from July 2018 to December 2022. Among them, forty-nine patients underwent bronchoalveolar lavage (BAL). Finally, 42 CIP patients with intact BALF analysis data were enrolled in our study (Fig. 1 shows the study flowchart). Follow-up information was obtained through outpatient follow-up records or telephone conversations with patients or their families. There were 4 kinds of outcomes in CIP patients, namely, cured, improved, deteriorated or dead.

Fig. 1
figure 1

Study flowchart

Full size image

Definitions

The diagnostic criteria for CIP were as follows [14]: 1) new onset or exaggeration of respiratory manifestations, especially dry cough, dyspnea, or a decrease in oxygen saturation (easily measured by a finger pulse oxygen saturation detector); 2) a positive history of immunotherapy with anti-PD-1/PD-L1; 3) new onset of lung shadows characterized by computed tomography (CT) features of ILD [15]; and 4) exclusion of infectious pulmonary diseases, primary lung cancer progression and/or metastatic malignancies, pulmonary edema, pulmonary hemorrhage, and other mimics.

The severity of CIPs was classified by the National Cancer Institute’s Common Terminology Criteria for Adverse Events (CTCAE, version 5.0) [16] when they were diagnosed with CIP: asymptomatic/mild (grade 1), moderate (grade 2), severe (grade 3), life-threatening (grade 4), or death (grade 5). In our study, patients with grade 1 or grade 2 CIPs were defined as low-grade CIPs, and patients with grade 3, grade 4 or grade 5 CIPs were defined as high-grade CIPs.

Serum cytokine levels were measured using IMMULITE 1000 Automated Immunoassay Analyzer (Siemens).

For BALF analyses, the following tests were performed: differential cell count, lymphocyte subset analysis, microbiological and virological tests, and malignant cell cytological examination. A minimal volume of 10 ml (generally 10 to 20 ml) of a pooled BALF sample was sent for BALF cellular analysis for each patient. The differential cell counts included macrophage, lymphocyte, neutrophil, and eosinophil cell counts.

T lymphocyte subset analysis of BALF was performed by flow cytometry. BALF samples were passed through a mesh filter to remove debris and then centrifuged to isolate cells. Cells were stained with a panel of flow antibodies included CD45-FITC, CD3-APC-A750, CD4-PC5.5, and CD8-APC-A700 (Beckman Coulter). Samples were analyzed using the Navios flow cytometry system.

Statistical analysis

The data were analyzed using the SAS version 9.4 software package (SAS Institute, Inc., SAS Campus Drive, Cary, North Carolina 27,513, USA). The GraphPad Prism 6XML project (GraphPad Software, San Diego, CA, USA) was used for graphing. Continuous variables are presented as means ± standard deviations (SD) for data with normal distribution, and as median and interquartile range (IQR) for non-normally distributed data. Normal distribution of continuous data was assessed with the Kolmogorov–Smirnov test. Categorical variables are expressed as frequencies and percentages. The t test or rank-sum test was used for continuous variables, and the chi-square test was used for categorical variables. A two-tailed P < 0.05 was considered to indicate statistical significance.

Results

Clinical characteristics of CIP patients

Forty-two CIP patients with complete BALF analysis data were enrolled in our study. There were 32 males (76.2%) and 10 females (23.8%). When they suffered from CIPs, they were 62.0 ± 10.4 (range: 31–78) years old. Most of them were older than 60 years (29 patients, 69.0%) and had a smoking history [26 patients/61.9%, 30 (20, 50) pack-year]. However, only 11 patients (26.2%) had emphysema. The range of Eastern Cooperative Oncology Group (ECOG) score was 0–2. Most of them were diagnosed with lung cancer (36 patients, 85.7%). There were 3 patients with esophageal squamous cell cancer, 1 patient with hepatic cell cancer, 1 patient with gastric signet ring cell cancer and 1 patient with pericardial malignant mesothelioma. The median time to onset of CIP was 98.5 days (3.28 months) after the first dose of ICIs.

According to the CTCAE, there were 2 patients (4.8%) with grade 1 CIPs, 16 patients (38.1%) with grade 2 CIPs, 20 patients (47.6%) with grade 3 CIPs and 4 patients (9.5%) with grade 4 CIPs. Therefore, there were 18 patients (42.9%) in the low-grade CIP group and 24 patients (57.1%) in the high-grade CIP group. The clinical and radiological characteristics of the patients in the low-grade CIP group and high-grade CIP group are listed in Table 1. There were no significant differences in age, male patient percentage or smoking history between patients with high-grade CIPs and those with low-grade CIPs (64.5 years vs. 61.5 years, P = 0.18; 79.2% vs. 72.2%, P = 0.60; 66.7% vs. 58.8%, P = 0.61).

Table 1 Clinical and radiological characteristics between low-grade and high-grade CIPs
Full size table

Lung cancer was more common in patients with low-grade CIPs than in patients with high-grade CIPs (100% vs. 75%, P = 0.01), however, there was no significant difference in preexisting ILDs between these two groups (58.3% vs. 50%, P = 0.65). During ICI treatment, patients with low-grade CIPs were more commonly treated with molecular targeted therapy than patients with high-grade CIPs (27.8% vs. 4.2%, P = 0.03). There was no significant difference in the prevalence of thoracic radiotherapy or chemotherapy combined with ICI treatment between patients with high-grade CIPs and those with low-grade CIPs (37.5% vs. 38.9%, P = 0.93; 79.2% vs. 83.3%, P = 0.59, respectively). Both organizing pneumonia (OP) and nonspecific interstitial pneumonia (NSIP) high-resolution CT (HRCT) patterns were the main radiological patterns for patients with high-grade and low-grade CIPs (41.7% vs. 61.1%, 37.5% vs. 33.3%, respectively; P = 0.13). Emphysema was more common on HRCT of patients with high-grade CIPs (37.5% vs. 11.1%, P = 0.046). Approximately 1/4 of the CIP patients in our study suffered from other irAEs concomitantly: 25% in the high-grade CIP group versus. 27.8% in the low-grade CIP group.

BALF cellular analysis

In our CIP cohort, the BALF differential cell counts are shown in Fig. 2, and the lymphocyte subset analysis is shown in Fig. 3. The BALF total cell count was 17.1 (11.9, 31.0) × 106/L, the macrophage percentage was 47.7 ± 23.9% of total live cells, the lymphocyte percentage was 36.7 ± 22.5%, the neutrophil percentage was 5.3% (2%, 16.4%), and the eosinophil percentage was 1% (0, 2.5%). A differential BALF cell count with greater than 15% lymphocytes was defined as a lymphocytic cellular pattern. Thirty-four CIP patients (81%) exhibited a lymphocytic cellular pattern.

Fig. 2
figure 2

The bronchoalveolar lavage fluid cell differential count results of checkpoint inhibitor pneumonitis (CIP) patients. A Total cell count of BALF. Data for individual patients as well as the median and interquartile range are shown. B Cell percentages of different cells. Data are presented using means and standard deviations

Full size image
Fig. 3
figure 3

The bronchoalveolar lavage fluid lymphocyte subset analysis results of checkpoint inhibitor pneumonitis (CIP) patients. A Lymphocyte percentage of CD3+, CD3+CD4+, and CD3+CD8+ cells. Data are presented using means and standard deviations. B CD4 + /CD8 + ratio. Data for individual patients as well as the median and interquartile range are shown. C Representative flow cytometric analysis of T cells in BALF

Full size image

As for the BALF lymphocyte subset analysis, the median percentage of CD3+ lymphocytes was 96.7% (94.3%, 97.6%) of total lymphocytes, the mean percentage of CD3+CD4+ lymphocytes was 31.0 ± 16.9% of total lymphocytes, the mean percentage of CD3+CD8+ lymphocytes was 58.8 ± 19.4% of total lymphocytes, and the median ratio of CD3+CD4+/CD3+CD8+ lymphocytes was 0.5 (0.3, 1.0). There were 31 CIP patients (73.8%) with a CD3+CD4+/CD3+CD8+ lymphocyte ratio less than 1; in other words, CD8+ lymphocytes were the main group of BALF T lymphocytes.

The BALF differential cell counts and lymphocyte subset analyses between high-grade and low-grade CIPs are shown in Table 2, and there was no significant difference between the two groups. There were 19 patients (79.2%) with high-grade CIP and 15 patients (83.3%) with low-grade CIP manifesting a lymphocytic cellular pattern. There were 18 patients (75%) with high-grade CIP and 13 patients (72.2%) with low-grade CIP having a CD3+CD4+/CD3+CD8+ lymphocyte ratio less than 1.0.

Table 2 Bronchoalveolar lavage fluid analysis between high-grade and low-grade checkpoint inhibitor pneumonitis (CIP) groups
Full size table

Serum cytokine analysis

There were 28 patients who underwent serum cytokine analysis at the time of diagnosis of CIP, including interleukin (IL)-6, IL-8, IL-10 and tumor necrosis factor (TNF)-α (Fig. 4). The serum concentrations of IL-6, IL-8, IL-10 and TNF-α were 21.7 ± 19.1 pg/ml (range: 2–77 pg/ml; normal reference range: < 5.9 pg/ml; 21 patients/75% with elevated serum IL-6), 28.1 ± 22.6 pg/ml (range: 7–98 pg/ml; normal reference range: < 62 pg/ml; 2 patients/7.1% with elevated serum IL-8), 9.1 ± 10.8 pg/ml (range: 5–51.9 pg/ml; normal reference range: < 9.1 pg/ml; 5 patients/17.9% with elevated serum IL-10), and 14.8 ± 14.9 pg/ml (range: 3–63.9 pg/ml; normal reference range: < 8.1 pg/ml; 19 patients/67.9% with elevated serum TNF-α).

Fig. 4
figure 4

The serum cytokine analysis results of checkpoint inhibitor pneumonitis (CIP) patients. The normal reference ranges for IL-6, IL-8, IL-10 and TNF-α were < 5.9 pg/ml, < 62 pg/ml, < 9.1 pg/ml, and < 8.1 pg/ml, respectively

Full size image

Discussion

The percentage of lymphocytes in the BALF was usually less than 15% in healthy individuals (8.4% in nonsmoker-healthy individuals and 3.8% in smoker-healthy individuals) [17, 18]. A BALF lymphocyte percentage greater than 15% was defined as a lymphocytic cellular pattern [17]. A lymphocyte percentage ≥ 25% was considered to indicate pulmonary granulomatous diseases, drug-related ILD, cryptogenic OP, NSIP or hypersensitivity pneumonitis (HP) [17]. Suresh et al. compared the BALF features of patients who were prescribed anticancer ICI therapy. The characteristics of the BALF differential cell count in patients without CIP after ICI administration were similar to those of healthy controls. However, the lymphocyte percentage in BALF increases as patients suffer from CIPs [9, 19]. In our cohort, the percentage of BALF lymphocytes was 36.7 ± 22.5%, and 34 CIP patients (81%) exhibited a lymphocytic cellular pattern.

Meyer et al. conducted a study to characterize the BALF T lymphocyte subset in clinically healthy volunteers in two different age groups (19–36) years (younger group) versus. 64–83 years (older group). The percentage of lymphocytes in the BALF fluid in the older group was greater than that in the younger group [17% vs. 8.3%, P < 0.05; 95% CI (2.7–14.7)]. The BALF CD3+CD4+/CD3 + CD8 + cell ratio was greater in the older group than in the younger group [(7.6 ± 1.5) vs. (1.9 ± 0.2), P < 0.001] [20]. There was no established normal reference range of BALF CD3+CD4+% or CD3+CD8+% for healthy individuals, and there were no special clinical indications for testing for BALF CD3+CD4+% and/or CD3+CD8+%. However, the BALF CD3+CD4+/CD3+CD8+ ratio seems to be correlated with several special ILDs, especially for pulmonary granulomatous diseases. The average percentages of BALF CD3+CD4+/CD3+CD8+ cells were 1.5 for smokers-healthy patients and 2.6 for nonsmokers-healthy patients [18]. For patients with ILDs, the presence of a BALF lymphocytic cellular pattern with an inverted (< 1.0) CD3+CD4+/CD3+CD8 + ratio could be a diagnostic indicator for HP. The BALF CD3+CD4+/CD3+CD8+ ratio in the HP ranged from 0.5 to 1.5 [18]. For most pulmonary sarcoidosis patients, BALF cell count differentials also exhibited a lymphocytic cellular pattern; however, BALF lymphocyte subset analysis revealed a high CD3+CD4+/CD3+CD8+ ratio. A BALF CD3+CD4+/CD3+CD8+ ratio ≥ 4 was strongly correlated with sarcoidosis and might be a diagnostic indicator for ILD patients with a BALF lymphocytic cellular pattern [17, 21, 22].

Suresh et al. [9] performed flow cytometry analysis of BALF from patients with CIP. In their study, both the BALF CD3+CD4+ and CD3+CD8+ T-cell percentages were approximately 10% of live cells. PD-1/PD-L1 inhibitors facilitate the activation of cytotoxic T cells against tumor cells [23, 24]. CD8+ T cells are the effectors of ICI antitumor effects and are associated with varied levels and states of CD8+ T-cell infiltration in tumors before and during ICI therapy [25]. Although the pathogenesis of CIP has not been identified until recently, it seems to be related to the dysregulated activation of T cells, especially CD8+ T cells, in the lung. In Suzuki’s CIP cohort, the percentage of BALF lymphocytes was 34.8% (7.5%-75%), and the ratio of BALF CD4+/CD8+ T-cell was 1.05 (0.07–4.33) [10]. Pulmonary tissues obtained from CIP and non-CIP NSCLC patients after ICI therapy by surgical biopsy were analyzed by Lin’s method: the number of CD8+ T cells was increased, while the number of resting CD4+ T cells was decreased in the CIP group [26]. Hone Lopez et al. performed immunohistochemical analysis of colon biopsies from patients with checkpoint inhibitor-induced colitis [7]. There was heavy mucosal infiltration of CD8+ T cells in the affected colon, both in the deep and superficial colon mucosa. CD8+ T cells also play a role in ICI-related Stevens-Johnson syndrome, toxic epidermal necrosis and vitiligo [27]. In our cohort, the BALF differential cell count also indicated lymphocytosis. Noteworthily, the median ratio of BALF CD3+CD4+/CD3+CD8+ lymphocytes was 0.5 (0.3, 1.0) in our study, which indicated a CD8+-dominant T cell inflammation. Suzuki et al. [10] also found that the proportion of BALF PD-1+ PD-L1+CD8+ T cells was correlated with the CTCAE grade of CIPs. However, there was no significant difference in the BALF T cell subset analysis between the high-grade and low-grade CIP groups in our study.

Serum proinflammatory cytokine levels were elevated in patients with irAEs, especially in those with severe irAEs. Elevated circulating and/or pulmonary inflammatory cytokines were reported in patients with CIP [19, 28]. IL-6 is an essential cytokine for the generation of Th17 cells from naïve CD4 + T cells [29], contributing to the development of several inflammatory/autoimmune diseases [30]. IL-6 is also a prognostic factor with higher serum levels associated with shorter overall survival in patients with metastatic melanoma receiving ICI in large randomized trials [31]. Serum levels of IL-6 at baseline and changes after immunotherapy might also have a prognostic significance in patients with advanced cutaneous squamous cell carcinoma [32]. It has been suggested that chronic inflammation and IL-6 cytokine signaling are involved in resistance to immunotherapy [33]. Targeting IL-6 could be an effective approach to treat irAEs without hindering antitumor immunity [34]. Compared with healthy controls, the BALF IL-6 concentration was elevated in patients with CIP; however, the serum IL-6 concentration was not elevated in Kowalski’s study [19]. Compared with those at baseline, both the serum and BALF IL-17A and IL-35 levels were elevated in patients with non-small cell lung cancer when they were suffering from CIP [28]. The level of BALF IFN-γ-induced protein 10 (IP-10) was elevated, and the level of IL-2 was decreased in patients with CIP compared with patients with pulmonary infection or lung cancer [35]. Given that inflammatory cytokines are not routinely tested in clinical practice, we could only analyze the serum inflammatory cytokine profiles of patients with CIPs. It was shown that the serum IL-6 concentration was significantly elevated in cancer patients who were suffering from CIPs. Tocilizumab, siltuximab, and JAK inhibitors are effective anti-IL-6 medications. Tocilizumab was suggested for severe irAEs in clinical practice [30]. JAK inhibitors were reported to be effective at treating fulminant myocarditis or steroid-resistant ICI-related acute transverse myelitis [4, 36]. Because T-cell activation is one of the potential pathogeneses of CIPs and because JAK inhibitors can effectively reduce T-cell activation [37], JAK inhibitors were assumed to effectively ameliorate severe CIPs.

This study has several limitations. First, not all CIP patients underwent BAL when they were suffering from CIPs, which might influence the exploration of differences between high-grade and low-grade CIPs because of the small sample size. Second, lung biopsies were not performed simultaneously with BAL fluid. Immunohistochemical analysis and/or single-cell sequencing of pulmonary tissues might be useful for identifying the key effector cells involved in CIPs. Third, there was no further exploration of the different states or subtypes of CD8+ T cells in our retrospective study. Fourth, as in most studies, there was no BALF T-cell subset analysis result for patients without CIPs who were administered anticancer ICI therapy, because it was not a routine test for these patients. We had no idea about the T-cell landscape for them. These patients should be included in our future well-designed prospective studies.

Conclusions

The CD3+CD8+ lymphocytosis pattern was the main inflammatory profile in the BALF of the CIP patients in our cohort. However, there was no significant difference in BALF features between patients with low-grade CIPs and those with high-grade CIPs. Targeting CD3+CD8+ lymphocytes might be a treatment option for CIPs.