This study was designed as a prospective observational cohort study conducted at a university-affiliated tertiary hospital in Korea. Eligible participants were adults aged 19 years or older, diagnosed with histologically confirmed NSCLC, and scheduled to receive PD-1/PD-L1 inhibitors. The study received approval from the Institutional Review Board of Korea University Guro Hospital (2022GR0010), and written informed consent was obtained from all participants. Although the original plan was to enroll patients over a 9-month period starting in February 2022, slow recruitment necessitated an extension of the inclusion period until June 2023.

Pneumonitis was clinically diagnosed based on the following criteria: the presence of newly developed pulmonary infiltrates during PD-1/PD-L1 inhibitor therapy, not fully attributable to infectious lung diseases or pulmonary edema. All clinically diagnosed pneumonitis events occurring during PD-1/PD-L1 inhibitor therapy were included in the analysis. Some patients had previously received thoracic radiotherapy. Distinguishing radiation-induced pneumonitis from immune-related pneumonitis was not always feasible. Therefore, all pneumonitis cases were analyzed regardless of etiology. The severity of pneumonitis was evaluated using the National Cancer Institute Common Terminology Criteria for Adverse Events Pneumonitis Grading System version 5.0 [12].

Chest computed tomography (CT) studies were independently reviewed, and the radiological patterns were classified by a radiologist with over 20 years of experience in thoracic imaging.

Serum KL-6 levels were assessed using the Nanopia KL-6 assay (Sekisui Medical, Tokyo, Japan). Baseline KL-6 levels were obtained from blood samples collected just before starting PD-1/PD-L1 inhibitor therapy. For patients who developed pneumonitis, KL-6 measurements were scheduled every 1-2 weeks from the onset of pneumonitis (onset KL-6) until the completion of management. In contrast, for patients who did not experience pneumonitis, KL-6 levels were planned to be measured every 6 weeks for up to 4 months. These measurement intervals were designed to align with real-world clinical practice, where patients receiving anticancer therapy typically undergo CT scans and blood tests every 1-2 months, with more frequent evaluations during episodes of pneumonitis.

Patients were divided into two groups: those with pneumonitis and those without. Categorical variables are presented as numbers and percentages, with group comparisons analyzed using Fisher’s exact test. Continuous data are shown as medians with interquartile ranges and compared using the Mann-Whitney U test. To assess differences in the longitudinal trajectory of KL-6 between the groups, we fitted a linear mixed-effects model to the log-transformed KL-6 values. This model included both a random intercept and a random slope for the time. Analyses were conducted using SPSS version 20.0 (IBM Corp., Armonk, NY, USA) and SAS version 9.4 (SAS Institute Inc., Cary, NC, USA). All statistical tests were two-sided, with a significance level set at p<0.05.

The median ages were 69 years for the pneumonitis group and 73 years for the non-pneumonitis group. All patients in the pneumonitis group (n=11) were male, while five of the seven patients in the non-pneumonitis group were male. The prevalence of smoking history and underlying lung disease was similar in both groups. In the pneumonitis cohort, squamous cell carcinoma was the predominant histologic type (n=6, 54.5%), whereas adenocarcinoma was more common in the non-pneumonitis group (n=5, 71.4%). Stage III disease was observed more frequently in the pneumonitis group (n=7, 63.6%), while stage IV disease was more prevalent in the non-pneumonitis group (n=6, 85.7%). The diffusion capacity of the lungs for carbon monoxide was numerically higher in the pneumonitis group (61% vs. 53%, p=0.791). Although baseline KL-6 levels appeared higher in the non-pneumonitis group, this difference was not statistically significant (261 U/mL vs. 373 U/mL, p=0.375). Further baseline details are provided in Table 1.

Patients in the pneumonitis group were more likely to have received prior chemotherapy or radiotherapy, although these differences did not reach statistical significance (81.8% vs. 42.9%, p=0.141; 72.7% vs. 42.9%, p=0.332). In the pneumonitis group, the median duration from the latest chemotherapy or radiotherapy to the onset of pneumonitis was 126 and 121 days, respectively. Durvalumab was the most frequently used PD-1/PD-L1 inhibitor in the pneumonitis group (63.6%), while pembrolizumab was predominantly administered in the non-pneumonitis group (57.1%). In the non-pneumonitis group, two patients underwent combination chemotherapy with PD-1/PD-L1 inhibitors, whereas no patients in the pneumonitis group did (Table 2).

Changes in serum KL-6 levels exhibited distinct patterns between the two groups (Figure 2 and Supplementary Figure S1). In the pneumonitis group, serum KL-6 levels generally showed an upward trend. In contrast, the non-pneumonitis group did not exhibit a clear overall change, with some patients experiencing reductions while others showed increases. In the linear mixed-effects model, the time × group interaction for log-transformed KL-6 did not reach statistical significance (p=0.113). Additionally, an exploratory receiver operating characteristic analysis evaluating the predictive value of KL-6 changes for pneumonitis also failed to yield statistically significant results (Supplementary Figure S2). The KL-6 levels for each individual patient in both groups are detailed in Supplementary Tables S1, S2.

KL-6 measurements at the onset of pneumonitis were available for five of the 11 patients, with a median value of 412 U/mL. All these patients showed an increase in KL-6 compared to their baseline levels at pneumonitis onset. The median duration from the initiation of PD-1/PD-L1 inhibitor therapy to pneumonitis onset was 48 days. Among these patients, eight experienced mild to moderate pneumonitis, while three had severe or life-threatening cases. CT imaging patterns were analyzed in 10 patients. Five of these patients exhibited localized infiltrates and all had previously received concurrent chemoradiotherapy before PD-1/PD-L1 inhibitor treatment (durvalumab). Among this group, four had main clinically relevant cancer lesions located in the right upper lobe, left lower lobe, right upper lobe, and right middle lobe, respectively, with their pneumonitis infiltrates corresponding to these lobes. The remaining patient had mediastinal lymph nodes (left subaortic and lower paratracheal nodes) as the main cancer site, while localized pneumonitis infiltrates were observed in both upper lobes. The other five patients presented with diffuse infiltrates: two were classified as having an acute interstitial pneumonia pattern, one as an organizing pneumonia pattern, and two as a nonspecific interstitial pneumonia/organizing pneumonia pattern (Figure 3). Both patients with nonspecific interstitial pneumonia/organizing pneumonia pattern had previously undergone radiotherapy. The main clinically relevant cancer lesions were located in the right lower lobe and left upper lobe, respectively. The patient with right lower lobe lesion showed infiltrates affecting the entire right lung, while the other with left upper lobe lesion primarily exhibited infiltrates in both lower lobes. Corticosteroid therapy was administered to seven patients, and four required hospitalization. One patient ultimately died after developing pneumonitis and undergoing mechanical ventilation. However, the primary cause of death was likely liver failure due to PD-1/PD-L1 inhibitor-related liver injury (Table 3).