Krebs von den Lungen-6 surveillance in immune checkpoint inhibitor-induced pneumonitis

Background

Pneumonitis induced on use of immune checkpoint inhibitors (ICIs) is defined as focal or diffuse inflammation of the lung parenchyma and is often associated with diminished survival rates.1 The onset of ICI-induced pneumonitis is believed to be influenced by lymphocyte activation, dysregulation of immunosuppressive cells, autoinflammatory cytokine cascades, and the presence of autoantibodies; however, the precise mechanisms remain unclear.2 Severe pneumonitis often leads to discontinuation of cancer treatment, requiring a transition to best supportive care.3 Consequently, impeding pneumonitis progression is paramount, highlighting the importance of effective screening and surveillance tools in clinical practice.

Current guidelines for immune-related adverse event (irAE) management advocate baseline CT before ICI administration and pulmonary function tests (PFTs) to assess pre-existing interstitial and chronic lung diseases, respectively.4–6 Surveillance strategies involve continuously monitoring symptoms such as dyspnea, persistent cough, chest pain, and fever and assessing oxygen saturation. In cases where pneumonitis is suspected, high-resolution CT and bronchoscopy with biopsy or bronchoalveolar lavage (BAL) sampling is recommended.4–6

Some patients are clinically asymptomatic at pneumonitis diagnosis.7 Chest radiography is insensitive in detecting pneumonitis, particularly in milder cases, and the presence of concomitant pulmonary disease complicates diagnosis through imaging tests.8 Therefore, a biomarker to objectively evaluate irAE pneumonitis is essential. Blood biomarkers offer distinct advantages of reproducibility, minimal invasiveness, and ease in analyzing. We previously demonstrated a gradual elevation in the neutrophil-to-lymphocyte ratio preceding pneumonitis onset.9 However, a common limitation of these serum biomarkers is their lack of specificity in pneumonitis, as they may be influenced by other irAEs.

Krebs von den Lungen-6 (KL-6), a sialylated glycoprotein discovered in 1985, is released into the bloodstream following lung injury and is known to reflect lung tissue damage and the regeneration of alveolar type II cells.10 Reportedly, serum and BAL levels of KL-6 are elevated in interstitial pneumonitis.11 Subsequent investigations in Asian and European populations demonstrated KL-6 elevation in various pulmonary conditions, including idiopathic pulmonary fibrosis (IPF), connective tissue disease-associated interstitial lung disease (CTD-ILD), and chronic hypersensitive pneumonitis.12–17 Additionally, Ohnishi et al identified a correlation between drug-induced pneumonitis and increased circulating KL-6 levels.18 Given this evidence, KL-6 is listed by experts as a potential blood test for drug-induced ILD, alongside surfactant protein (SP)-A and SP-D.19 In the context of ICI-induced pneumonitis, which leads to pulmonary damage similar to other etiologies, KL-6 levels could increase and previous study findings underscore the possibility of KL-6 as a serum biomarker in pneumonitis.

In this study, we retrospectively analyzed historical KL-6 levels in patients undergoing ICI treatment and scrutinized their fluctuations during pneumonitis across diverse cancer types. Our primary objective was to assess the potential clinical feasibility of KL-6 as a screening biomarker to identify ICI-induced pneumonitis.

MethodsPatient selection

This retrospective study involved patients diagnosed with non-small cell lung cancer (NSCLC), malignant melanoma, head and neck cancer (HNC), renal cell carcinoma (RCC), hepatocellular carcinoma (HCC), esophageal cancer, gastric cancer, urothelial cancer, and microsatellite instability (MSI)—high solid cancer. These patients received ICIs as part of standard therapy at Kyushu University Hospital, Japan, between September 2014 and March 2021. Patients who lacked baseline KL-6 measurements before commencing ICI treatment or had not undergone KL-6 assessment after treatment initiation were excluded. Follow-up data were available for all patients through March 31, 2022.

Diagnosis and investigation of ICI-induced pneumonitis

ICI-induced pneumonitis was initially diagnosed by the treating oncologist and subsequently confirmed by a multidisciplinary team, which included specialists from oncology, hematology, gastroenterology, endocrinology, pulmonology, nephrology, cardio-oncology, dermatology, neurology, hepatology, as well as oncology nurses and pharmacists. The diagnosis was based on clinical and radiographic evidence, carefully excluding infectious pneumonia and disease progression. Additionally, BAL was conducted when necessary to assess lymphocyte percentage or CD4/CD8 ratio. Pneumonitis severity was graded using the National Cancer Institute Common Terminology Criteria for Adverse Events V.5.0. To evaluate the effect of patient background characteristics on the diagnostic utility of KL-6, we investigated several factors, including smoking history, history of asthma or chronic obstructive pulmonary disease (COPD), presence of lung metastasis (Group 1 only), history of prior lung radiation, history of ILD, history of prior chemotherapy associated with ILD risk, history of tyrosine kinase inhibitor treatment (Group 2 only), and history of concomitant medication with ILD risk. Patients with chemotherapy or medications with ILD risk were defined as those having received drugs reported to cause ILD, as outlined previously.19 Pneumonitis classification with respect to steroid responsiveness was based on the criteria established previously.20 Briefly, patients showing clinical improvement within 3 days of appropriate steroid therapy were classified as having steroid-responsive pneumonitis, while those not as having steroid-unresponsive pneumonitis.

Measurement of KL-6 and SP-A and SP-D

Serum KL-6 concentration (U/mL) was measured using a chemiluminescent enzyme immunoassay (CLEIA) by Lumipulse Presto KL-6 (Sekisui Medical, Tokyo, Japan). SP-A (ng/mL) and SP-D (ng/mL) measurements using a CLEIA were outsourced at a clinical testing company (LSI Medience, Tokyo, Japan).

Statistical analysis

Clinical variables dichotomized by pneumonitis were compared using Fisher’s exact test. Factors affected by baseline KL-6 values were evaluated using multivariate regression analysis. Data between the two groups were compared using the Mann-Whitney U test and Wilcoxon matched-pairs test, and among more than three groups using the Kruskal-Wallis test with post hoc Dunn’s multiple comparison test. The diagnostic performance of KL-6 was evaluated using receiver operating characteristic (ROC) curve analysis, and the area under the curve (AUC) was calculated to quantify the overall ability of KL-6 to diagnose pneumonitis. The optimal cut-off value was determined using the Youden index. All statistical analyses were performed using GraphPad Prism V.10.2.3 (GraphPad Software, La Jolla, California, USA) and JMP Pro V.17.0.0 (SAS Institute, Cary, North Carolina, USA). All tests were two-sided; p values<0.05 indicated statistical significance.

Ethical declarations

This study was conducted following the Declaration of Helsinki.

ResultsPatient classification and prevalence of pneumonitis

We conducted a study involving 500 patients who received ICI treatment for adjuvant, advanced, or metastatic disease stages. Both univariate and multivariate analyses identified patient with NSCLC status as the sole independent factor linked to elevated baseline KL-6 levels. (Online supplemental figure 1a–f), (online supplemental table 1). Based on these findings, we defined the following groups: Patients with melanoma, HNC, RCC, HCC, gastric cancer, esophagus cancer, urothelial cancer, and MSI-high cancer were classified as Group 1 (n=382) and patients with NSCLC as Group 2 (n=118) (figure 1). In Group 1, 37 patients (9.7%) developed pneumonitis, with 11 cases of grade≥3 pneumonitis. In Group 2, 24 patients (20.3%) developed pneumonitis, with eight cases of grade≥3 (figure 1). Group 2 had significantly higher baseline KL-6 levels than that of Group 1 (Group 1: 221.5 U/mL, Group 2: 408.0 U/mL, p<0.0001, (online supplemental figure 1g). The median duration of pneumonitis onset was 2.4 (0.1–34.2) months. There were no differences in patient backgrounds between those who developed pneumonitis and those who did not in each group (online supplemental table 2 and 3).

Figure 1Figure 1Figure 1

Research flow diagram for this study. HCC, hepatocellular carcinoma; HNC, head and neck cancer; irAE, immune-related adverse event; MSI, microsatellite instability; NSCLC, non-small cell lung cancer; RCC, renal cell carcinoma.

Diagnostic significance of KL-6 in pneumonitis

In Group 1, KL-6 levels were significantly elevated at pneumonitis onset (pre: 222.0 U/mL, post: 743.0 U/mL, p<0.0001, figure 2a). KL-6 increase was observed regardless of pneumonitis severity (figure 2b). ROC analysis showed an AUC of 0.903 (sensitivity 81.1%, specificity 91.6%) and a cut-off value 1.52 times the baseline value (figure 2c). In Group 2, KL-6 levels were elevated (pre: 360.5 U/mL, post: 506.5 U/mL, p=0.029, figure 2d), with no correlation with pneumonitis severity (figure 2e). ROC analysis yielded an AUC of 0.683 (sensitivity 45.8%, specificity 87.2%, figure 2f). Since KL-6 was initially established as a biomarker for interstitial pneumonitis, a comparison of its diagnostic feasibility across several etiologies is shown in table 1.

Table 1

KL-6 as a diagnostic marker: comparison with previous studies

Figure 2Figure 2Figure 2

Diagnostic performance of KL-6 for pneumonitis. (a) Comparison of KL-6 (U/mL) levels at pretreatment and pneumonitis onset in cancers other than non-small cell lung cancer (Group 1). For patients without pneumonitis, KL-6 levels were measured at around 2.4 months, the median time of pneumonitis onset. (b) Correlation of KL-6 elevation with pneumonitis severity. (c) Receiver operating characteristic (ROC) curve analysis for evaluating the diagnostic performance of KL-6 for pneumonitis. (d–f) The same analyses as in (a–c) were performed in Group 2. Data are presented as median with IQR. Significance was determined using the Wilcoxon test (a, d) and Kruskal-Wallis test with post hoc Dunn’s multiple comparison test (b, e). In the ROC analysis, the Youden index was employed to determine the optimal cut-off value (c, f). AUC, area under the curve; irAE, immune-related adverse event; KL-6, Krebs von den Lungen-6.

Influence of patient factors on the diagnostic significance of KL-6

To further evaluate the diagnostic utility of KL-6, we investigated the potential influence of various patient factors, including smoking history, history of asthma or COPD, lung metastasis, prior lung radiation, history of ILD, and the use of ILD-risk drugs, such as tyrosine kinase inhibitors. In patients with non-NSCLC (Group 1), baseline KL-6 levels were higher in smokers, patients with a history of ILD, and those who had undergone prior chemotherapy associated with ILD risk (online supplemental figure 2). However, these factors did not influence the pneumonitis incidence or the elevation of KL-6 levels at pneumonitis onset (online supplemental table 2 and Figure 3). Among patients with NSCLC (Group 2), baseline KL-6 levels were affected only in those with adenocarcinoma histology (online supplemental figure 4). ROC analysis revealed an AUC of 0.694 (80.0% sensitivity, 59.3% specificity) in patients with adenocarcinoma and an AUC of 0.536 (50.0% sensitivity, 76.2% specificity) in patients with squamous cell carcinoma, indicating that KL-6 was less reliable in NSCLC regardless of the histology (online supplemental figure 5). KL-6 elevation was observed at pneumonitis onset, irrespective of the presence of concurrent irAEs, and KL-6 levels remained unchanged during the onset of non-pulmonary irAEs (online supplemental figure 6 and 7).

SP-A and SP-D levels at pneumonitis onset: correlation with KL-6

SP-A and SP-D levels were measured at the onset of irAE pneumonitis in a subset of patients, and their diagnostic accuracy was evaluated in those for whom data were available. The percentage of patients with SP-A levels above the reference value (43.8 ng/mL) was 68.8% in Group 1 (11 of 16 patients) and 42.9% in Group 2 (six of 14 patients) (online supplemental figure 8a). Regarding SP-D levels above the reference value (110.0 ng/mL), it was 54.5% in Group 1 (12 of 22 patients) and 58.8% in Group 2 (10 of 17 patients) (online supplemental figure 8b). In Group 1, where KL-6 was a reliable biomarker for pneumonitis, there was no significant correlation between KL-6 and elevated SP-A or SP-D (SP-A: r=0.406, p=0.134; SP-D: r=0.113, p=0.622; Online supplemental figure 8c and d).

KL-6 elevation in symptomatic and asymptomatic pneumonitis

In Group 1, we assessed the presence or absence of symptoms at pneumonitis onset. Of 37 patients, 19 were asymptomatic, while 18 exhibited symptoms. The condition of 5 patients worsened over time; 14 patients were ultimately classified as having grade 1 and 23 as having grade≥2 pneumonitis (figure 1). At pneumonitis onset, KL-6 levels were significantly elevated in both symptomatic (pre: 205.0 U/mL, post: 674.5 U/mL, p<0.0001) and asymptomatic patients (pre: 314.0 U/mL, post: 743.0 U/mL, p<0.0001) (figure 3a). There was no significant difference in KL-6 elevation from baseline values, regardless of subjective symptoms (figure 3b).

Figure 3Figure 3Figure 3

KL-6 elevation in symptomatic and asymptomatic pneumonitis. (a) Comparison of KL-6 (U/mL) levels at pretreatment and pneumonitis onset in patients with or without subjective symptoms among cancers other than non-small cell lung cancer (Group 1). (b) Comparison of KL-6 elevation ratio with or without subjective symptoms at pneumonitis onset. Data are presented as median with IQR. Significance was determined using the Wilcoxon test (a) and Mann-Whitney U test (b). KL-6, Krebs von den Lungen-6; irAE, immune-related adverse event; NS, not significant.

KL-6 fluctuations reflect steroid responsiveness in pneumonitis

We evaluated KL-6 levels to assess the responsiveness of pneumonitis to steroid therapy. For Group 1 patients who received steroids, we recorded their KL-6 levels before and after treatment initiation. Patients were classified as either steroid-responsive (n=8; four received methylprednisolone (mPSL) pulse therapy, three received 0.8–1.0 mg/kg prednisolone (PSL) equivalent, and one received 0.5 mg/kg PSL equivalent) or steroid-unresponsive (n=6; four received mPSL pulse therapy, and two received 0.5 mg/kg PSL equivalent). In the steroid-responsive group, KL-6 levels either remained stable or exhibited a slight decrease after 1 month (from 1030.0 to 855.0 U/mL, p=0.547, (online supplemental figure 9a). In the steroid-unresponsive group, KL-6 levels increased significantly 1-month post-treatment (from 678.0 to 1078.0 U/mL, p=0.031, (online supplemental figure 9b).

Discussion/conclusion

Here, we investigated the clinical utility of KL-6 as a surveillance marker for pneumonitis. KL-6 significantly increased in patients with various cancer types other than NSCLC, with high sensitivity and specificity comparable with those reported in previous ILD studies. The KL-6 increase was observed regardless of the severity and presence of subjective symptoms at onset. KL-6 was increased irrespective of underlying lung conditions such as smoking, lung metastasis, and ILD history. Furthermore, KL-6 did not change at the development of non-pulmonary irAEs, suggesting high specificity for ICI-induced pneumonitis. To the best of our knowledge, this is the first study to report the clinical feasibility of KL-6 as a highly specific serum marker for ICI-induced pneumonitis.

Although KL-6 was elevated when pneumonitis occurred in patients with NSCLC, it could not accurately detect pneumonitis. Reportedly, the diagnostic accuracy of KL-6 was low in patients with lung cancer who developed pneumonitis, consistent with our results.21 For patients with NSCLC, current guidelines recommend early detection of pneumonitis by monitoring subjective symptoms and continuing imaging studies.4–6

As shown in table 1, the diagnostic accuracy of KL-6 in IPF and CTD-ILD has been reported in various Asian and European countries. Lee et al reported an AUC of 0.860 (sensitivity 79.4%, specificity 79.9%) in CTD-ILD.12 Zheng et al reported an AUC of 0.911 (sensitivity 85.3%, specificity 90.0%) in idiopathic interstitial pneumonia.15 Ohnishi et al reported a sensitivity of 93.9% and specificity of 96.3% compared with those for healthy people or those with bacterial pneumonia.17 A previous meta-analysis reported an AUC of 0.88 (sensitivity 76.0%, specificity 89.0%) in CTD-ILD.22 The only report evaluating KL-6 for drug-induced lung injury showed a sensitivity of 53.3%.18 Based on previous studies, the clinical feasibility of KL-6 as a biomarker for pneumonitis in this study (AUC 0.903, sensitivity 81.1%, specificity 91.6%, Group 1) appears to be as accurate as its reported efficacy in other etiologies. The cut-off value for KL-6 was calculated using the relative increase from the baseline value (1.52 times the baseline value). While absolute values are clinically easier to use, in this study, they were inferior in sensitivity and specificity than relative values (AUC 0.822, sensitivity 78.4%, specificity 81.7%, cut-off value 400.0 U/mL).

KL-6 levels were elevated even during mild or asymptomatic pneumonitis, suggesting that KL-6 monitoring helps detect pneumonitis early, potentially reducing severe or fatal cases. However, KL-6 levels did not correlate with the severity of pneumonitis and cannot help assess disease activity. KL-6 levels correlate with disease activity in CTD-ILD, as assessed using semiquantitative CT grading and lower PFTs.12 Further studies are needed to scrutinize the correlation of KL-6 variability with high-resolution CT evaluation or PFT at pneumonitis onset.

Apart from KL-6, SP-A and SP-D are also considered serum biomarkers for pneumonitis.19 Our results showed that SP-A and SP-D were less sensitive than KL-6 at pneumonitis onset. Reportedly, KL-6 had the highest diagnostic accuracy compared with SP-A, SP-D, and monocyte chemoattractant protein-1 in interstitial pneumonitis.17 Additionally, KL-6 was found to be superior to SP-A in diagnosing ILD,16 and SP-D has lower sensitivity and specificity than KL-6 in differentiating CTD-ILD.22 These findings align with our results, suggesting that KL-6 is the most reliable marker for pneumonitis.

KL-6 was initially discovered through experiments targeting lung cancer-specific antigens. Baseline KL-6 levels were elevated in patients with NSCLC and exceeded the cut-off for pneumonitis in Group 1 patients. Reportedly, KL-6 correlates with anticancer treatment response in patients with NSCLC.23 Therefore, fluctuations in KL-6 levels due to histology and treatment response or disease progression after ICI treatment initiation may hinder the detection of KL-6 elevation associated with pneumonitis in patients with NSCLC.

This study has some limitations. First, this being a single-center study, validation in multicenter settings or larger cohorts is required. To further establish the utility of KL-6 as a screening tool for ICI-induced pneumonitis, false negatives need to be investigated, as seen in patients with NSCLC in this study. Although non-NSCLC cancers were grouped together in this analysis, future studies with larger cohorts should analyze each cancer type individually. We must also consider the influence of chemotherapy or molecular-targeted agents used in combination and perform further analysis based on each regimen. This remains a key focus of our future research. Second, previous reports indicate baseline KL-6 levels differ among Japanese, Chinese, and German populations, suggesting racial differences.24 Our results need confirmation in other racial groups. Third, the correlation between KL-6 elevation and ILD activity evaluated using CT needs further research. Because pneumonitis exhibits various histopathological (eg, organizing pneumonia, non-specific interstitial pneumonia, diffuse alveolar damage) and radiological features (eg, cryptogenic organizing pneumonia-like patterns, ground-glass opacities, interstitial pneumonia, hypersensitivity pneumonitis, and pneumonitis not otherwise specified),1 it is essential to determine if our results apply to each pneumonitis subtype.

In conclusion, this study showed that monitoring KL-6 levels during ICI therapy was reliable for detecting pneumonitis in patients with conditions other than NSCLC. However, in patients with NSCLC, the significance of KL-6 monitoring is limited as KL-6 elevation is less sensitive for irAE detection. Early detection through KL-6 monitoring allows for timely intervention, preventing treatment interruptions, and minimizing the use of immunosuppressants. Further evidence needs to be accumulated, and if the results are validated in a large cohort and the biomarker can be routinely measured in clinical practice with insurance coverage, it could become a valuable screening tool for a broader range of ICI-treated patients. We believe our study will enhance the management of pneumonitis and improve patient care for those undergoing ICI therapy.

留言 (0)

沒有登入
gif