Changes in soluble PD-L1 levels in patients with advanced non-small cell lung cancer

Between May 2018 and October 2022, the cross-sectional study (N = 110) enrolled 80 patients diagnosed with advanced-stage NSCLC and 30 healthy volunteers (control group). The control group consisted of individuals undergoing routine health check-ups at the Clinic of Military Hospital 103. Inclusion criteria for the lung cancer group involved newly diagnosed patients with stage IIIB and IV NSCLC with PS 0–2. Exclusion criteria included concurrent cancers, immune-compromising conditions, and history of use of immune-suppressing drugs including systemic corticosteroids. Similar criteria were applied to the control group, emphasizing regular health check-ups and excluding individuals with cancer or immune-related conditions. Detailed clinical assessments, radiologic examinations, diagnostic biopsies, and immunohistochemical analyses were conducted, alongside the examination of mPD-L1 expression in tissue samples and sPD-L1 concentration in serum. The normal control group underwent blood tests, chest X-rays, and sPD-L1 concentration testing in serum. Lung cancer staging followed the 7th edition TNM classification by the Union for International Cancer Control (UICC), with the histological type determined using the World Health Organization (WHO) classification of lung tumors (2015)^[13]. The study received ethical approval from the Institutional Review Board of Military Hospital 103 (No. 99B/15-Aug-2022/VMMU-IRB), and the two groups provided informed consent.

mPD-L1 expression testing utilizes immunohistochemical techniques with a single PD-L1 antibody [rabbit anti-human monoclonal antibody PD-L1 (clone 73-10) from Leica, USA], performed on the Leica BOND-MAX automated staining system. The intensity of PD-L1 staining is assessed on a scale from 0 to 2+ as follows: 0 (negative): no staining or faint staining in less than 1% of tumor cells; 1+ (low): weak staining in 1%–49% of tumor cells; and 2+ (strong): strong staining in more than 50% of tumor cells. The standard requires a minimum of 100 cancer cells^[14]. We used the Tumor Proportion Score (TPS) as a critical metric used to quantify PD-L1 expression in tumor cells. It serves as a key determinant in selecting patients for certain immunotherapies, particularly those targeting the PD-1/PD-L1 pathway. Both positive and negative control samples are used to validate the specificity and sensitivity of the PD-L1 antibody.

Serum was obtained by centrifugation (1300×g for 10 min) and then aliquoted and stored at −80 °C until study analysis. sPD-L1 concentration testing employs a pre-assembled enzyme-linked immunosorbent assay (ELISA) kit with a specific PD-L1 antibody (SEA788Hu) from Cloud-Clone Corporation (USA)^[15]. The linearity was validated by testing serial dilutions of known sPD-L1 concentrations, achieving an acceptable R² value of greater than 0.99. The assay demonstrated a detection limit of 0.1 ng/mL, validated by spiking known amounts of sPD-L1 into sample matrices and measuring the lowest detectable signal above the background. The assay specificity was confirmed through cross-reactivity and interference testing, ensuring selective detection of sPD-L1. Intra-assay and inter-assay precision were validated, with acceptable coefficients of variation (CVs) of less than 10%. In addition, sPD-L1 concentrations were normalized to total protein concentration measured by a bicinchoninic acid (BCA) assay, allowing the expression of sPD-L1 levels relative to total protein content in the samples.

In the statistical analysis, data were processed using the medical statistics software SATA 15.0. Categorical variables were delineated by frequencies and proportions, while normally distributed continuous variables were presented using mean and standard deviation and non-normally distributed ones were described by median and interquartile range. The Mann–Whitney test was used to compare medians for variables lacking a normal distribution. Receiver operating characteristic (ROC) curve analysis was conducted to calculate the area under the curve (AUC) and determine the optimal cutoff point for sPD-L1, optimizing sensitivity and specificity. AUC ≥0.60: poor accuracy; ≥0.70: average accuracy, ≥0.80: good accuracy; and ≥0.90: very good accuracy. The cutoff value is selected at the point with the highest Youden index (J) with J = Se + Sp 1, and the threshold value is determined with the “Cutpt” and “Roctab” commands in Stata 15.0 software. Univariate logistic regression analysis chooses a testing threshold of p < 0.2 to select variables, combining characteristics based on clinical experience to include in the multivariable logistic regression model, to predict the likelihood of occurrence, indicates high or low PD-L1 levels, the optimal model is when most of the subjects are accounted for by the model and (in this study, the number of subjects satisfying this model is 67 patients) Pseudo R² >20%, p < 0.05.

The majority of lung cancer patients were in stage IV (80%), presenting hazy masses in 66.25% and malignant pleural effusion in 47.5%. Adenocarcinoma (ADC) dominated histologically (72.5%), followed by SCC (20%) and ASC (7.5%). EGFR mutation occurred in 24.0% of cases. Comparing the lung cancer group to normal groups, the mean serum PD-L1 concentration was significantly higher (1.08 vs. 0.56 ng/mL and 1.08 vs. 0.42 ng/mL, respectively; p < 0.001). mPD-L1 expression was observed in 61.2% of patients, with 28.75% having high expression and 32.5% having low expression. Group demographics showed no significant age or body mass index (BMI) differences, but the control group had a lower male ratio (p < 0.001), with nonsmokers in the control group. sPD-L1 concentrations did not significantly differ by gender, age group, smoking history, disease stage, tumor size, or histological type (p > 0.05). The mean sPD-L1 value in lung cancer patients with mPD-L1 expression was 1.09 ng/mL, which was slightly higher than that in the non-mPD-L1 group (0.93 ng/mL), with no statistically significant difference (p = 0.3). No correlation was found between sPD-L1 and mPD-L1 expression levels (p = 0.726).

Comparing the study group to the control group, the diagnostic threshold for sPD-L1 was determined at 0.92 ng/mL, with 56.25% sensitivity and 83.33% specificity [AUC = 0.756; 95% confidence interval (CI): 0.663–0.849; p = 0.001]. Multivariate regression analysis indicated associations between increased sPD-L1 and factors like age, disease onset duration, tumor size, and finger clubbing, although statistical significance was not reached for most factors (p > 0.05). The model demonstrated statistical significance overall (p = 0.031), suggesting a potential role for sPD-L1 in diagnosing advanced NSCLC.

Table 1 outlines the common characteristics of the study group. The findings align with previous research and selection criteria for this study. The patients were typically older, predominantly male, and with a smoking history, larger tumor size, ADC histology, and advanced-stage disease. Gender differences between the study group and the control group were influenced by smoking habits in Vietnam, where the majority of smokers are male, a primary risk factor for NSCLC. In addition, gender differences were observed between the UTP group and the normal control group. Despite limited relevance in previous studies, gender differences were maintained in our control group^[16]. Previous studies also reported the highest proportion of ADC in NSCLC^[17].

In this study, 49 patients (61.25%) exhibited mPD-L1 expression, with a median of 10 ng/mL. Among these, 23 patients (46.93%) showed high expression. The positive rate of mPD-L1 varied across studies, subjects, and the type of immunotherapy used [Table 5]. Despite using different PD-L1 antibodies, our study’s TPS findings were concordant, with thresholds set at ≥1% (positive) and <1% (negative). The TPS ≥1% subgroup was further divided at cutoffs ≥50% and <50%. Our study, using the PD-L1 73-10 antibody, correlated with the clinical trials JAVELIN Lung 100 and JAVELIN Lung 200, reporting positive mPD-L1 rates of 56.4% and 66%, respectively, while our study found a positive rate of 61.25%^[29,30]. Generally, mPD-L1 expression rates were consistently above 60% across studies [Table 5].

sPD-L1, a soluble form generated by cleavage or splicing of PD-L1 mRNA, circulates in the serum of lung cancer, autoimmune, and viral infection patients^[33,34]. Our study demonstrated significantly higher sPD-L1 concentrations compared to both control groups (p < 0.01) [Table 3]. This aligns with previous research showing elevated sPD-L1 levels in cancer patients^[11]. The diagnostic ROC curve analysis revealed an optimal AUC of 0.817 when compared to the normal control group. The diagnostic cutoff for UTP was 0.92 ng/mL, providing 56.25% sensitivity and 77.33% specificity. Thus, sPD-L1 could serve as a promising prognostic indicator for lung cancer alongside other biomarkers^[34,40]. However, the prognostic value of sPD-L1 in immune therapy response was not evaluated in our study. Multivariate regression analysis indicated associations between elevated sPD-L1 levels and older age, prolonged disease duration, larger tumor size, and finger clubbing, though statistical significance was not reached (p > 0.05). Further exploration with a larger sample size is warranted to validate these findings and improve model utility^[14,37].

Figure 1:

ROC curve of serum PD-L

Figure 2:

Distribution characteristics of mPD-L1 and sPD-L1.