Predictive Role of Baseline Peripheral Lung SUVmax on PET/CT for Immune-Related Pneumonitis and Adverse Events in Lung Cancer Patients Treated with Immune Checkpoint Inhibitors
Article information
Abstract
Purpose
Immune checkpoint inhibitors (ICIs) offer durable responses in lung cancer patients lacking actionable mutations or with resistance to prior therapies. However, predicting their efficacy and associated immune-related adverse events (irAEs), such as severe pneumonitis, remains a clinical challenge. This study investigated the predictive value of positron emission tomography/computed tomography (PET/CT)–derived metabolic parameters for pneumonitis and other irAEs in lung cancer patients treated with ICIs.
Materials and Methods
We retrospectively analyzed 151 patients with advanced non–small cell lung cancer (NSCLC) who received ICIs as first-line treatment, either as monotherapy or in combination. Pre-treatment PET/CT was used to measure maximum standardized uptake value (SUVmax) at the primary tumor, tumor-uninvolved peripheral lung, and metastatic sites. Pneumonitis and irAEs were assessed using clinical and radiological findings.
Results
Pneumonitis, severe pneumonitis (grade ≥ 3), and irAEs occurred in 26.5%, 19.9%, and 37.1% of patients, respectively. A peripheral SUVmax cutoff of > 1.1 significantly predicted pneumonitis (area under the curve, 0.720; p < 0.001). High peripheral SUVmax was associated with higher rates of pneumonitis (42.9% vs. 12.3%, p < 0.001), severe pneumonitis (31.4% vs. 9.9%, p=0.001), and irAEs (46.4% vs. 29.5%, p=0.038). In multivariate analysis, high peripheral SUVmax independently predicted pneumonitis (odds ratio [OR], 4.621; 95% confidence interval [CI], 1.868 to 11.431; p=0.001), severe pneumonitis (OR, 2.848; 95% CI, 1.043 to 7.779; p=0.041), and irAEs (OR, 2.509; 95% CI, 1.114 to 5.504; p=0.022).
Conclusion
Baseline peripheral SUVmax on PET/CT may serve as a noninvasive biomarker for predicting immune-related pneumonitis and other irAEs in NSCLC patients receiving ICIs, supporting early risk identification.
Introduction
With the introduction of immune checkpoint inhibitors (ICIs) in lung cancer treatment, new therapeutic options have become available for patients with refractory lung cancer or without targetable mutations [1,2]. However, a significant limitation of ICIs in lung cancer therapy is the lack of reliable biomarkers for predicting their efficacy and immune-related adverse events (irAEs). Although extensive research has focused on programmed cell death ligand 1 (PD-L1) expression and tumor mutation burden, these factors are insufficient predictors of ICI outcomes. Some patients with high PD-L1 expression may exhibit a minimal response to ICIs, whereas patients without PD-L1 expression may experience favorable outcomes [2]. Furthermore, unique tumor responses associated with ICIs, such as hyperprogression and pseudoprogression, underscore the substantial unmet need for predictive biomarkers [3]. Additionally, ICIs are associated with a distinct spectrum of irAEs, among which pneumonitis is one of the most serious and potentially life-threatening [2]. Compared to patients with other solid tumors, those with lung cancer develop pneumonitis more frequently and often with more severe clinical manifestations, likely due to coexisting pulmonary comorbidities such as chronic obstructive pulmonary disease (COPD), emphysema, or subclinical interstitial changes [4]. Immune-related pneumonitis may necessitate treatment interruption or discontinuation and, in severe cases, can lead to respiratory failure and death [5]. Therefore, identifying patients at increased risk of pneumonitis prior to initiating ICI therapy remains a critical unmet need in clinical practice.
Previous studies have explored various clinical and radiographic risk factors for ICI-related pneumonitis, including a history of interstitial lung disease (ILD), COPD, prior thoracic radiation therapy, and smoking [6]. While these factors may offer some insight into the underlying lung vulnerability, they often rely on morphological interpretation of computed tomography (CT) images and may not fully capture the dynamic immune microenvironment or subclinical inflammation underlying pneumonitis pathogenesis [7]. Moreover, many of these structural abnormalities do not consistently predict pneumonitis in diverse patient populations, and there is a lack of objective, quantitative imaging biomarkers that can be used to guide risk stratification before treatment initiation.
Positron emission tomography/computed tomography (PET/CT), routinely performed during initial staging and treatment planning for lung cancer, offers the opportunity to noninvasively assess pulmonary metabolic activity [8]. In particular, fluorodeoxyglucose (FDG) uptake in the lung parenchyma, measured using standardized uptake values (SUV), may reflect ongoing immune activation, as well as subclinical or potential inflammation [9]. Maximum SUV (SUVmax) refers to the highest radiotracer uptake observed within a region of interest on PET imaging, reflecting the area of greatest metabolic activity. High SUVmax values on PET suggest areas of elevated glucose metabolism, which may reflect tumor aggressiveness, active inflammation, or increased immune activity. Although tumor-related PET parameters have been widely studied for prognostic purposes [10], relatively little is known about the predictive value of non-tumoral PET metrics, especially those from the peripheral lungs, in forecasting irAEs such as pneumonitis. Given the growing interest in leveraging metabolic and inflammatory PET imaging to capture early immunologic changes, PET-derived parameters may serve as novel biomarkers for pneumonitis susceptibility.
In this study, we aimed to investigate whether baseline metabolic parameters obtained from PET/CT imaging could serve as useful predictors for the development of immune-related pneumonitis and adverse events in patients with non–small cell lung cancer (NSCLC) receiving ICIs as first-line treatment. We also explored how these metabolic imaging features, compared with conventional structural lung abnormalities, such as ILD and emphysema, are commonly evaluated on chest CT.
Materials and Methods
1. Patients and treatment
This retrospective cohort study included patients with advanced NSCLC who received first-line treatment with pembrolizumab between June 2017 and February 2024 at Chungnam National University Hospital (CNUH). Eligible patients had baseline PET/CT images obtained before initiating pembrolizumab therapy. Patients were treated with pembrolizumab either as a monotherapy or in combination with cytotoxic chemotherapy, in accordance with the treating physician’s discretion and clinical guidelines. Treatment was continued until the patient experienced serious adverse events, confirmed investigator-assessed disease progression, or withdrew informed consent. Patients deemed by the investigator to be experiencing a clinical benefit could continue treatment beyond radiological disease progression.
2. Assessment of underlying lung abnormalities on chest CT
Baseline chest CT findings were recorded, with the most recent scan obtained before ICI treatment initiation considered as the baseline. Radiographic features suggestive of ILD or emphysema were evaluated. ILD was identified based on the presence of reticular opacities, ground-glass attenuation, or honeycombing. Emphysema was defined as low-attenuation areas without defined walls, typically in the upper lobes. CT images were independently assessed by a board-certified thoracic radiologist with over 20 years of experience, who was blinded to the clinical outcomes.
3. 18F-FDG PET/CT acquisition
18F-FDG PET/CT was performed using the Discovery 690 PET/CT system (GE Medical Systems). Patients were instructed to fast for 4-6 hours prior to the examination. A dose of 18F-FDG (3.7 MBq/kg body weight) was injected intravenously, and the examination was performed after a 1-hour rest. The administration of 18F-FDG was contraindicated in cases where the blood glucose level exceeded 200 mg/dL. Attenuation correction of positron emission tomography (PET) images using CT data was performed. CT was performed using a continuous spiral 64-slice technique with a tube tension of 120 kV, pitch of 1.0, slice thickness of 3.75 mm, and rotation speed of 0.5 s/revolution. The scan coverage included the area from the skull base to the mid-thigh.
For acquisitions used in photon attenuation and scatter correction, a fixed tube current of 50 mA·s was used; otherwise, tube current modulation was applied. To enable image fusion and the generation of CT-based transmission maps, CT data were downscaled from a 512×512 matrix to a 128×128 matrix to match the PET resolution. PET image reconstruction was performed using attenuation correction based on CT data, employing an ordered subset expectation maximization iterative algorithm. Images were displayed and analyzed using an Advanced Workstation (GE Medical System) with multiplanar reformatted images.
4. 18F-FDG PET/CT image analysis
SUVmax values were measured for the primary tumor, peripheral lung parenchyma, and the most avid metastatic lesions, denoted as SUVmax (cancer), SUVmax (peripheral), and SUVmax (metastatic site), respectively. SUV was calculated as follows: SUV=(decay-corrected activity in kBq/mL of tissue volume)/(injected 18F-FDG activity in kBq/body weighting).
Peripheral SUVmax was defined as the highest FDG uptake observed in non-malignant peripheral lung parenchyma. In principle, measurements were obtained from the lung contralateral to the primary tumor to avoid peritumoral or tumor-associated inflammatory uptake. However, if a distinct region of elevated uptake, unrelated to the tumor, was observed in the ipsilateral lung (e.g., consistent with underlying ILD or inflammation distant from the tumor), that region was used instead. Regions adjacent to the tumor or likely to represent tumor-induced inflammation were excluded from analysis (Fig. 1).
Measurement of maximum standardized uptake value (SUVmax) in primary tumor and peripheral lung parenchyma. (A-C) Images of a patient with a primary lung tumor (yellow arrow) in the left upper lobe (LUL): (A) computed tomography (CT), (B) positron emission tomography (PET), and (C) PET/CT fusion image. The SUVmax was measured at the primary tumor (yellow arrow) and at the contralateral peripheral lung parenchyma (blue arrow) to define the peripheral SUVmax. (D-F) In cases where there is prominent uptake in the ipsilateral lung, such as in underlying lung disease, the peripheral SUVmax was measured from the area showing the highest uptake in the same lung. (D) CT image of an LUL lung cancer. (E, F) CT and PET/CT fusion images showing increased fluorodeoxyglucose uptake in the underlying lung disease in the left lower lobe. The yellow arrow indicates the primary tumor, and the blue arrow indicates the area of increased uptake in the ipsilateral peripheral lung parenchyma, from which the peripheral SUVmax was measured.
5. Treatment response and irAEs
The treatment response was evaluated using CT scans at regular intervals during treatment. For patients receiving pembrolizumab monotherapy, response assessments were performed every three treatment cycles. For those receiving pembrolizumab in combination with cytotoxic chemotherapy, CT evaluation was conducted every two cycles during the initial four treatment cycles, followed by every three cycles thereafter. Treatment response was assessed according to the Response Evaluation Criteria in Solid Tumors (RECIST) ver. 1.1. Clinical benefit was defined by the disease control rate, comprising both partial response (PR) and stable disease (SD).
IrAEs were defined as toxicities resulting from immune system dysregulation, potentially affecting various organs, including the skin, gastrointestinal tract, liver, endocrine glands, and lungs. The grade of irAEs was defined using the Common Terminology Criteria for Adverse Events (CTCAE) ver. 5.0. Severe irAEs were defined as ≥ grade 3.
6. Diagnosis of immune-related pneumonitis
The diagnosis of immune-related pneumonitis was established based on the following criteria: (1) new or worsening respiratory symptoms occurring during or after ICI treatment, (2) newly developed bilateral ground-glass opacities or consolidations on chest CT, (3) absence of pulmonary infection, confirmed by negative microbiological testing and/or lack of clinical response to empirical antibiotics, and (4) exclusion of alternative etiologies such as cardiac dysfunction (e.g., heart failure). According to the CTCAE ver. 5.0, pneumonitis grading was defined as follows: grade 1, asymptomatic; grade 2, symptomatic, medical intervention indicated, limiting instrumental activities of daily living (ADL); grade 3, severe symptoms, limiting self-care ADL, oxygen indicated; grade 4, life-threatening respiratory compromise requiring urgent intervention; and grade 5, death. Severe pneumonitis in this study was defined as grade ≥ 3.
7. Statistical analysis
To calculate the sensitivity and specificity of biomarkers, conventional receiver operating characteristic (ROC) curves were generated, and the area under the curve (AUC) was calculated. The optimal cutoff value was determined as the point at which the Youden index was maximized by the ROC curve. Data are expressed as means±standard deviation. Chi-squared and independent t tests were used to analyze differences in patients’ clinicopathological data. Pearson’s correlation coefficients were calculated to analyze relationships among variables. Univariate logistic regression analyses were performed to identify factors associated with pneumonitis, severe pneumonitis, and irAEs. Variables with a p-value < 0.1 in univariate analyses were subsequently included in multivariate logistic regression models to adjust for potential confounders. Odds ratios (ORs) with 95% confidence intervals (CIs) were calculated, and a p-value < 0.05 was considered statistically significant. SPSS ver. 22 (IBM Corp.) and MedCalc ver. 23.2.1 (MedCalc Software Ltd.) were used for all statistical analyses.
Results
1. Patient baseline characteristics
The baseline demographic and clinical characteristics of the 151 patients included in this study are summarized in Table 1. The mean age was 71.8±9.7 years, and most patients were male (82.8%) and had a history of smoking (72.8%). Most patients were diagnosed with advanced-stage disease, with 33.8% having stage IVA disease and 65.6% having stage IVB disease. The predominant histologic subtype was adenocarcinoma (56.3%), followed by squamous cell carcinoma (37.1%). PD-L1 tumor proportion scores (TPS) were distributed as follows: TPS < 1% in 43.0%, 1%-49% in 15.9%, and ≥ 50% in 40.4% of patients. ICI treatment consisted of immuno-oncology (IO) monotherapy in 28.5% of patients and an IO-chemotherapy combination in 71.5%. Comorbid emphysema and ILD were present in 41.1% and 19.2% of patients, respectively. Treatment responses included PR in 38.4%, SD in 38.4%, and progressive disease in 16.6%. IrAEs of any grade occurred in 37.1% of patients, with grade ≥ 3 irAEs reported in 22.5%. A total of 81 irAE events were observed, as some patients experienced more than one toxicity. Pneumonitis was the most frequently reported irAE, followed by cutaneous toxicities such as drug eruption and rash, and then hepatitis. A detailed breakdown of irAE types and frequencies is provided in S1 Table. Pneumonitis developed in 26.5% of patients, and severe pneumonitis (grade ≥ 3) occurred in 19.9%. Mean baseline SUVmax values were 14.8±6.7 for primary tumors, 1.8±1.7 for peripheral lung regions, and 12.1±7.1 for metastatic sites.
Baseline characteristics of the study population
2. Factors associated with the development of pneumonitis and severe pneumonitis
A comparison of baseline characteristics between patients with and without pneumonitis is summarized in Table 2. There were no significant differences in age, sex, smoking status, clinical stage, histology, PD-L1 expression, or treatment regimens between the two groups. Emphysema was more common among patients who developed pneumonitis than those who did not (52.5% vs. 36.9%, p=0.086). Importantly, ILD was significantly more common in patients who developed pneumonitis (32.5% vs. 14.4%, p=0.013). Additionally, baseline peripheral SUVmax values were significantly higher in the pneumonitis group compared to the no pneumonitis group (2.7±2.3 vs. 1.5±1.3, p=0.003).
Baseline patient characteristics according to the presence of all grade pneumonitis and severe pneumonitis
When comparing patients with severe pneumonitis to those with no or mild-to-moderate pneumonitis, emphysema (63.3% vs. 35.5%, p=0.006) and ILD (36.7% vs. 14.9%, p=0.007) were significantly more common in the severe pneumonitis group. Furthermore, peripheral SUVmax was significantly higher in patients who developed severe pneumonitis compared to those with no or mild-to-moderate pneumonitis (3.0±2.5 vs. 1.5±1.3, p=0.003).
3. Predictive value of baseline peripheral SUVmax for pneumonitis and severe pneumonitis occurrence
We focused on peripheral rather than central SUVmax because underlying ILD typically manifests in peripheral lung regions, and immune-related pneumonitis tends to be more severe in these areas [11].
The distribution of baseline peripheral SUVmax values among all patients is shown in Fig. 2A. Most patients had relatively low peripheral SUVmax values, with a small subset exhibiting markedly elevated values. The predictive performance of baseline peripheral SUVmax for pneumonitis development was evaluated using ROC curve analysis (Fig. 2B). The optimal cutoff value for peripheral SUVmax was determined to be > 1.1, with an AUC of 0.720, demonstrating significant predictive accuracy (p < 0.001). Based on this cutoff, patients were stratified into low and high peripheral SUVmax groups. As shown in Fig. 2C, the incidence of pneumonitis was significantly higher in the high SUVmax group compared to the low SUVmax group (42.9% vs. 12.3%, p < 0.001). Similarly, the incidence of severe pneumonitis was higher in the high SUVmax group than in the low SUVmax group (31.4% vs. 9.9%, respectively; p=0.001).
Predictive value of baseline peripheral maximum standardized uptake value (SUVmax) for pneumonitis and severe pneumonitis occurrence. (A) Scatter plot showing the distribution of baseline peripheral SUVmax values across all patients. (B) Receiver operating characteristic curve for the prediction of pneumonitis using peripheral SUVmax values. The optimal cutoff value was > 1.1, with an area under the curve (AUC) of 0.720 and a statistically significant predictive performance (p < 0.001). (C) Incidence of pneumonitis and severe pneumonitis according to low and high peripheral SUVmax groups. Patients with high peripheral SUVmax showed significantly higher rates of pneumonitis (p < 0.001) and severe pneumonitis (p=0.001) compared to those with low peripheral SUVmax.
Representative cases further illustrate the predictive relevance of the peripheral lung SUVmax. As shown in Fig. 3, one patient with combined pulmonary fibrosis and emphysema (CPFE) and low baseline peripheral SUVmax did not develop pneumonitis despite the presence of underlying lung disease, whereas another patient with idiopathic pulmonary fibrosis (IPF) and high peripheral SUVmax experienced severe pneumonitis following ICI therapy (Fig. 3).
Positron emission tomography/computed tomography and chest computed tomography findings according to peripheral lung f luorodeoxyglucose uptake and pneumonitis development. (A-C) A patient with squamous cell carcinoma combined with pulmonary f ibrosis and emphysema received cytotoxic chemotherapy combined with immune checkpoint inhibitor (immuno-oncology [IO]) therapy. Despite the presence of underlying lung disease, the patient had a low baseline peripheral lung maximum standardized uptake value (SUVmax) (blue circle in B) and showed a partial tumor response without evidence of pneumonitis. (D-F) A patient with squamous cell carcinoma and underlying idiopathic pulmonary fibrosis (IPF) who received immune checkpoint inhibitor (IO) monotherapy. This patient exhibited a high peripheral lung SUVmax (blue circle in E) at baseline, and follow-up computed tomography revealed new multifocal ground-glass opacities in both lungs, consistent with severe immune-related pneumonitis. CPFE, combined pulmonary fibrosis and emphy sema.
4. Multivariate analysis of risk factors for pneumonitis and severe pneumonitis
The results of univariate and multivariate logistic regression analyses for the development of pneumonitis and severe pneumonitis are summarized in Table 3. In the univariate analysis, ILD (OR, 2.859; 95% CI, 1.225 to 6.672; p=0.015) and high peripheral SUVmax (> 1.1) (OR, 5.325; 95% CI, 2.360 to 12.016; p < 0.001) were significantly associated with pneumonitis. However, in the multivariate analysis, only high peripheral SUVmax remained an independent predictor of pneumonitis (OR, 4.621; 95% CI, 1.868 to 11.431; p=0.001). For severe pneumonitis, univariate analysis identified emphysema (OR, 3.133; 95% CI, 1.365 to 7.190; p=0.007), ILD (OR, 3.313; 95% CI, 1.353 to 8.113; p=0.009), and high peripheral SUVmax (OR, 4.182; 95% CI, 1.722 to 10.158; p=0.002) as significant factors. In the multivariate model, emphysema (OR, 2.504; 95% CI, 1.054 to 5.948; p=0.038) and high peripheral SUVmax (OR, 2.848; 95% CI, 1.043 to 7.779; p=0.041) were independently associated with severe pneumonitis risk.
Univariate and multivariate analyses of factors associated with the development of pneumonitis and severe pneumonitis
5. Association between baseline characteristics and the development of irAEs
Baseline characteristics of patients according to the occurrence of irAEs are summarized in Table 4. No significant differences were observed between patients with and without irAEs regarding age, sex, smoking status, clinical stage, histological subtype, or PD-L1 expression status. Patients who developed irAEs were more likely to have received ICI monotherapy compared to those who did not (39.3% vs. 22.1%, p=0.024). Additionally, the incidence of ILD was significantly higher in the irAE group (30.4% vs. 12.6%, p=0.008). Regarding metabolic parameters, the baseline peripheral SUVmax was significantly higher in patients who developed irAEs than in those who did not (2.4±2.1 vs. 1.5±1.3, p=0.003). There were no significant differences in baseline primary tumor SUVmax or metastatic site SUVmax between the two groups.
Baseline patient characteristics according to irAEs
6. Risk factors for the development of irAEs
Univariate and multivariate logistic regression analyses for irAE development are presented in Table 5. In univariate analysis, ICI monotherapy (OR, 2.280; 95% CI, 1.107 to 4.697; p=0.025), ILD (OR, 3.015; 95% CI, 1.313 to 6.922; p=0.009), and high baseline peripheral SUVmax (> 1.1) (OR, 3.229; 95% CI, 1.621 to 6.432; p=0.001) were significantly associated with irAE development. In the multivariate analysis, only high baseline peripheral SUVmax remained an independent predictor of irAE development (OR, 2.509; 95% CI, 1.114 to 5.504; p=0.022), whereas ICI monotherapy and ILD were no longer significant.
Univariate and multivariate analyses of factors associated with the development of immune-related adverse events
Discussion
In this study, we demonstrated that the baseline peripheral lung SUVmax measured using PET/CT was significantly associated with the development of immune-related pneumonitis and other irAEs in patients with NSCLC receiving ICIs. A peripheral SUVmax cutoff of > 1.1 was predictive of pneumonitis with good discriminative performance (AUC, 0.720), and high peripheral SUVmax was identified as an independent predictor for both immune-related pneumonitis and irAEs in multivariate analyses. Furthermore, patients with elevated peripheral SUVmax values had significantly higher rates of severe pneumonitis (grade ≥ 3), suggesting its potential utility in identifying individuals at risk for clinically meaningful toxicities. To further explore the clinical implications of peripheral SUVmax in predicting pneumonitis, we compared baseline characteristics between patients with high and low peripheral SUVmax. Among previously reported risk factors—such as preexisting pulmonary comorbidities (particularly ILD and emphysema), smoking history, and prior thoracic radiation [12,13]—ILD and emphysema were more frequent in the high SUVmax group, but high SUVmax remained an independent predictor in multivariate analysis. Patients with prior radiation therapy were not included, and there were no differences between the two groups in smoking history, PD-L1 expression, or histology. Although patients in the high SUVmax group were more often older and male, neither age nor sex was significantly associated with pneumonitis development. To our knowledge, this is one of the first studies to evaluate PET/CT-derived non-tumoral lung parameters as predictive biomarkers of ICI-related toxicity in a real-world cohort.
The role of ICIs in lung cancer treatment is rapidly expanding [1]. ICIs, initially developed for metastatic lung cancer, are now integral to consolidation therapy after concurrent chemoradiotherapy in unresectable locally advanced lung cancer [14]. Their use has gradually extended to earlier stages in NSCLC, including neoadjuvant, adjuvant, and perioperative treatment [15]. Although numerous studies have investigated biomarkers for predicting ICI efficacy and the risk of irAEs, including pneumonitis, clinically reliable markers remain limited in real-world practice. Lung cancer patients often have comorbid lung diseases, including COPD and ILDs such as IPF and CPFE. These conditions share common risk factors like smoking and increase the risk of respiratory adverse events during ICI therapy, particularly severe pneumonitis [16,17]. Various clinical parameters have been proposed as potential predictors of irAEs, including sex, antibiotic exposure, post-treatment neutrophil-to-lymphocyte ratio, and baseline circulating tumor cell levels. In addition, the baseline prognostic nutritional index, body mass index, and post-treatment lactate dehydrogenase levels are associated with the severity of irAEs [18]. The presence of pleural or pericardial metastases has also been reported to increase the risk of ICI-related adverse events [19]. However, there remains a need for more objective and quantifiable biomarkers to improve risk stratification for immune-related toxicity.
FDG PET/CT is routinely used for staging lung cancer and plays a central role in baseline assessment and treatment planning [8]. Beyond its established use in tumor staging, recent evidence has highlighted the potential of PET/CT in evaluating immune-related complications and predicting treatment outcomes in patients receiving ICIs [20]. For example, FDG uptake patterns are associated with irAEs, such as thyroiditis, colitis, and pneumonitis, which can help differentiate toxicity from disease progression or infection [21,22]. In addition, harmonized PET-based response criteria such as PERCIST and immune modified PET Response Criteria in Solid Tumors (imPERCIST) have demonstrated prognostic relevance in ICI-treated NSCLC, with changes in SUVmax correlating with survival outcomes [23]. A growing number of studies have explored the prognostic value of tumor-derived PET parameters, including SUVmax, metabolic tumor volume (MTV), and total lesion glycolysis (TLG). Higher baseline SUVmax and MTV values have been associated with durable responses and improved survival in patients receiving anti–programmed death-1 therapy [24,25]. Longitudinal analyses further suggest that reductions in total MTV and TLG during treatment may reflect a favorable response and improved prognosis [26]. Meta-analytic data also support MTV as a reliable predictor of both overall and progression-free survival in patients undergoing ICI therapy, whereas other parameters, such as SUVmax or TLG, have been more variable [27]. Total metabolic tumor burden has been shown to predict clinical outcomes in patients receiving ICIs [28], and both MTV and TLG have been associated with survival in patients with high PD-L1 expression undergoing first-line immunotherapy [29,30]. While these studies underscore the value of PET-derived tumor metabolism as a predictive tool, the majority of the research to date has focused on tumor-centric metrics. In contrast, little is known about how non-tumorous lung metabolism, particularly in the peripheral parenchyma, may be related to the development of irAEs, such as pneumonitis. In our cohort, pneumonitis was the predominant irAE, accounting for 71.4% of patients who experienced irAEs, suggesting that the overall association between peripheral SUVmax and irAE incidence is largely driven by pneumonitis. However, after excluding pneumonitis cases, the incidence of non-pneumonitis irAEs remained higher in the high SUVmax group compared to the low SUVmax group (32.0% vs. 19.7%), although the difference did not reach statistical significance (p=0.118). Notably, mean peripheral SUVmax was significantly higher in patients who experienced non-pneumonitis irAEs than in those who did not (2.32±2.08 vs. 1.48±1.32, p=0.039). We hypothesize that elevated peripheral SUVmax may reflect subclinical immune activation in non-tumorous lung parenchyma, which could contribute to a broader spectrum of systemic irAEs beyond pneumonitis. Emerging evidence suggests that the lung is not only a target but also a source of systemic immune modulation; for example, lung-resident memory T cells have been shown to exit into systemic circulation under certain conditions and initiate immune responses in distant organs [31,32]. This raises the possibility that immune priming in the lungs—detected via increased metabolic activity on PET/CT—may serve as an early signal of a heightened systemic immunologic state, thereby predisposing patients to multiple forms of irAEs.
Several studies have investigated the clinical relevance of FDG uptake in non-tumorous lung parenchyma. Elevated FDG activity in morphologically normal lung tissue has been associated with pulmonary complications, including pneumonitis and acute exacerbation of ILD. For instance, the pre-treatment 90th percentile of SUV (SUV_P90) in non-tumorous lungs has been shown to correlate with early-onset radiation pneumonitis in patients receiving chemoradiotherapy followed by immunotherapy [33]. In patients with preexisting ILD, a higher baseline SUVmax in the lung parenchyma was associated with acute exacerbation following systemic chemotherapy [34]. Another study demonstrated that a higher peripheral lung-to-mediastinum SUVmax ratio (rSUVmax) predicted progressive ILD in lung cancer patients undergoing lobectomy [11]. Collectively, these findings suggest that non-tumor metabolic activity on PET/CT, particularly within the peripheral lung parenchyma, may provide clinically relevant information for identifying patients at risk of pulmonary complications. A recent study by Yamazaki et al. [35] also examined baseline lung FDG uptake as a predictor of ICI-related ILD, reporting conceptually similar findings. While their artificial intelligence–based volumetric segmentation of the entire noncancerous lung offers a sophisticated means to minimize interobserver variability, our approach demonstrates that comparable predictive value can be achieved using a simpler, clinically feasible metric—peripheral lung SUVmax—from routine PET/CT. This highlights the robustness and practicality of our method for real-world application, especially as it requires no specialized software or additional processing. Furthermore, our cohort was restricted to treatment-naïve advanced NSCLC patients receiving first-line ICI-based therapy, minimizing confounding from prior treatments. These methodological and clinical distinctions support the complementary rather than redundant nature of our findings. In addition, our study found that 22.9% of patients in the high peripheral SUVmax group had no radiologic evidence of underlying lung disease, 50% of them developed pneumonitis. This finding suggests that increased FDG uptake in morphologically normal lung tissue may reflect subclinical inflammation or immune priming, which predisposes patients to immune-related pulmonary toxicity following ICI therapy.
We investigated the clinical utility of peripheral lung FDG uptake on baseline PET/CT as a predictor of irAEs, particularly pneumonitis, in patients with NSCLC receiving ICIs. The incidence of immune-related pneumonitis in our study (26.5% overall; 19.9% grade ≥ 3) was higher than that reported in many real-world series (~5%-15%) [36]. This difference is likely attributable to the high proportion of high-risk patients in our cohort, including 19% with pre-existing ILD—nearly half of whom developed pneumonitis—and over 40% with emphysema, as well as a relatively older patient population. Unlike many clinical trials that exclude such high-risk individuals or focus primarily on severe symptomatic cases, we systematically identified pneumonitis through both clinical and radiologic surveillance, capturing even milder cases. This comprehensive approach may account for the higher observed incidence and underscores the need for vigilant monitoring in similar high-risk populations. Given these findings, this study underscores the potential value of integrating PET/CT-derived metrics into the personalized management of lung cancer patients. By leveraging routinely acquired imaging data, clinicians may be able to stratify patients according to risk and individualize treatment plans to mitigate immune-related toxicities. These findings have important clinical implications and demonstrate the feasibility of repurposing existing diagnostic tools to improve the safety and precision of ICI therapy. Multicenter validation and the incorporation of complementary clinical and molecular biomarkers will be essential for refining predictive models. Ultimately, this approach may help bridge the gap between routine clinical practice and precision oncology, thereby ensuring that ICIs are used safely and effectively in patients with lung cancer.
In conclusion, our findings suggest that baseline peripheral lung metabolic activity measured by PET/CT may serve as a non-invasive biomarker for predicting immune-related toxicity, particularly pneumonitis, in patients with NSCLC receiving ICIs. Integrating peripheral SUVmax into pre-treatment risk stratification may help identify high-risk patients, enabling more proactive monitoring and early clinical intervention.
Electronic Supplementary Material
Supplementary materials are available at Cancer Research and Treatment website (https://www.e-crt.org).
Notes
Ethical Statement
This study was conducted in accordance with the Declaration of Helsinki and Good Clinical Practice guidelines and approved by the institutional review board of our institution (IRB No. 2025-06-015 at CNUH). The requirement for informed consent was waived by the IRB board owing to the retrospective nature of the study.
Author Contributions
Conceived and designed the analysis: Kim KH, Kang DH, Chung C.
Collected the data: Kim KH, Kim SM, Lee JE, Kim SS, Kang DH, Chung C.
Contributed data or analysis tools: Kim KH, Kim SM, Chung C.
Performed the analysis: Kang DH.
Wrote the paper: Kim KH, Kang DH, Chung C.
Conflicts of Interest
Conflict of interest relevant to this article was not reported.
Funding
This work was supported by the National Research Foundation of Korea (NRF) grants funded by the Korean Government (MSIT) (grant numbers 2022R1A2C2010148 and RS-2022-NR071878). This research was supported by grants from the Korea Health Technology R&D Project through the Korea Health Industry Development Institute (KHIDI), funded by the Ministry of Health & Welfare, Republic of Korea (grant numbers RS-2020-KH088690 and RS-2022-KH130308). This work was supported by Chungnam National University Hospital Research Fund, 2024-CF-004. This work was supported by Chungnam National University.