PD-L1 expression correlation with metabolic parameters of FDG PET/CT and clinicopathological characteristics in non-small cell lung cancer

Lung cancer is the most common cancer and the leading cause of cancer death all over the world [1], and non-small cell lung cancer (NSCLC) is the most commonly diagnosed histological subtype [2‐4]. Despite the development in treatment for NSCLC such as thoracoscopic surgery, chemotherapy, radiotherapy, and targeted therapy [5, 6], the overall 5-year survival rate is still poor [7]. According to the data from 2009 to 2015, the 5-year survival rate of patients with lung cancer is as low as 19.4%, while that of patients with distant metastasis is only 5.2% [8]. Nowadays, many studies indicated that immunotherapy targeting programmed cell death ligand-1 (PD-L1) could provide a promising new way for the treatment of NSCLC [9, 10]. Antibodies blocking PD-1/PD-L1 pathway can enhance anti-tumor immunity to achieve the goal of killing cancer cells [11]. Thus, it is necessary to explore the existence of immune checkpoints and to select NSCLC candidates for immunotherapy.

At present, immunohistochemistry is the main method for identifying tumor PD-L1 expression, which requires surgical or biopsied tumor specimens from patients with NSCLC [12]. However, these procedures such as surgery, transbronchial lung biopsy, or CT-guided biopsy are invasive and even failed due to the patients’ physical condition or unqualified specimens. Fluorine-18 fluorodeoxyglucose (FDG) PET/CT is a noninvasive imaging modality and can be applied even if specimens from patients are not available, which has been widely applied to predict many molecular phenotypes of malignant tumors, such as histological types, tumor differentiation, proliferation, hypoxia, and genetic mutation [13‐17].

The previous studies have confirmed that tumor expression of PD-L1 was related to high glucose metabolism of PET/CT in NSCLC [18‐22]. However, almost all previous studies focused on the correlation of the FDG PET/CT metabolic parameter of maximum standard uptake (SUVmax) with PD-L1 expression [18‐22]. SUVmax is easy to determine and reflects the maximum uptake of glucose by metabolically active lesions. Volume-based PET parameters such as metabolic tumor volume (MTV) and total lesion glycolysis (TLG) represent glucose activity in the entire tumor mass, which could also reflect the metabolic status of the malignant tumors [23]. However, at present, little is known about the association of PD-L1 with MTV and TLG.

Therefore, in this study, FDG PET/CT findings and metabolic parameters of SUVmax, MTV and TLG of primary lesion (MTV-P and TLG-P), and combination of primary lesion and metastases (MTV-C and TLG-C) were retrospectively analyzed in 374 patients with NSCLC. We investigated tumor PD-L1 positive expression correlation with these metabolic parameters and patients’ clinicopathological characteristics and analyzed their roles in predicting PD-L1 expression in NSCLC.

In recent years, immunotherapy targeting PD-L1 has shown promising results in NSCLC patients. Different TPS values were used in previous studies for evaluating tumor PD-L1 expression, including 1%, 5%, 10%, and 50% [26, 27]. The latest version of the National Comprehensive Cancer Network reported NSCLC patients could benefit from immunotherapy when TPS was over 1% [12]. For example, when TPS ≥ 50%, PD-1/PD-L1 antibody combined with chemotherapy is the best choice for ACC, and PD-1/PD-L1 antibody alone is the best choice for SCC. When TPS during 1% to 49%, PD-1/PD-L1 antibody combined with chemotherapy is prior to PD-1/PD-L1 antibody alone for both ACC and SCC. When TPS < 1%, PD-1/PD-L1 antibody combined with chemotherapy is better than chemotherapy alone for both ACC and SCC [28]. Therefore, we chose TPS ≥ 1% as PD-L1 positive in this study and investigated PD-L1 expression correlation with metabolic parameters of FDG PET/CT and clinicopathological characteristics in NSCLC.

Several reports suggested that the association between the percentages of PD-L1 positive tumors and histological subtypes of NSCLC was found although the criteria of PD-L1 positivity varied in each study. Janzic et al. [26] reported that PD-L1 positive (TPS ≥ 5%) cases were higher in SCC (52%) than in AC (17%). Lin et al. [27] showed that PD-L1 positivity (TPS ≥ 1%) was more frequently observed in SCC (46%) than in AC (27%). Miyazawa et al. [29] compared the percentages of PD-L1 positive NSCLCs using the same criteria as Lin et al. [27], which were confirmed to be more frequent in SCC (44%) or large cell carcinoma (67%) than in AC (21%). In this study, the percentage of tumors with PD-L1 positive expression was 27.8%, 21.6%, and 47.3% in overall NSCLC, AC, and SCC, respectively. PD-L1 positive rate of SCC was significantly higher than that of AC, which was consistent with the previous reports [26, 30]. Moreover, our previous study [22] demonstrated that the positive rate of PD-L1 expression in pulmonary sarcomatoid carcinoma (PSC) was 87.5%, which was higher than that of SCC and AC in this study under the same criteria.

At present, some studies have demonstrated that the predictive value of SUVmax on FDG PET/CT in PD-L1 expression from the primary tumor in patients with lung cancer at the initial diagnosis. For example, Kaira et al. [21] demonstrated that PD-L1 expression level was significantly correlated with SUVmax in lung AC. Zhang et al. [19] reported a significant correlation between PD-L1 expression levels and SUVmax in pulmonary SCC. Takada et al. reported the association between PD-L1 expression and SUVmax in patients with small-sized lung cancer [18] or NSCLC [20], respectively. Our previous study also confirmed that SUVmax could be used to assess PD-L1 expression of PSC [22]. Similar to these previous reports, the results of the present study showed that PD-L1 expression was associated with SUVmax in overall NSCLC, lung AC, and SCC. It may be due to altered metabolism of tumor-infiltrating T lymphocytes resulting from tumor-imposed glucose restriction. Chang et al. [31] showed the glucose consumption by tumors metabolically restricts T cells, leading to their dampened mTOR activity, glycolytic capacity, and IFN-γ production, thereby allowing tumor progression. They also found that blocking PD-L1 directly on tumors dampens glycolysis by inhibiting mTOR activity and decreasing expression of glycolysis enzymes, reflecting a role for PD-L1 in tumor glucose utilization.

However, Jreige et al. [32] reported that no significant correlation was observed between PD-L1 tumor expression and following parameters such as SUV, MTV, and TLG in 49 cases with confirmed NSCLC. SUV represents the FDG activity of the tumor mass. MTV and TLG represent incorporating information on both tumor volume and metabolic factors. These metabolic parameters of PET/CT could reflect the glucose metabolic status of tumor tissues, which is an important feature of tumor biology [23]. Therefore, contrary to the results of Jreige et al. [32], our study found that in addition to SUVmax, TLG-P and TLG-C were also correlated with PD-L1 positivity in NSCLC. With regard to MTV, it represents the metabolic volume of the tumor mass, and different from the calculation of TLG, FDG activity is not included in the measurement of MTV. Our study showed that PD-L1 positivity had no association with the tumor metabolic volume as well as the tumor size. Thus, it demonstrated that the PD-L1-positive expression was mainly associated with FDG activity. Furthermore, when pathologically subgrouped, only SUVmax had the relationship with PD-L1 expression, which further displayed that the relationship between PD-L1 positivity and glucose activity of the tumor mass.

Moreover, this study also demonstrated that PD-L1 expression was significantly correlated with man, smoking ,and central NSCLC, which were also consistent with the previous studies [18, 20, 33]. It suggested that male, smoking-associated and central NSCLC such as lung SCC tended to exhibit PD-L1 positive expression, which confirmed the findings that PD-L1 positive rate of lung SCC was significantly higher than that of lung AC [26, 27, 29]. Furthermore, when the population of this study was pathologically subgrouped, the correlation between PD-L1 expression and these clinicopathological features disappeared, which was also in agreement with others’ reports [9, 34]. It was mainly due to that male, smoking and central tumor were major features of lung SCC, which were eliminated the interrelation with PD-L1 when compared intra-group. Moreover, the relationship between PD-L1 expression and EGFR mutation of NSCLC remains controversial. Several studies suggested that PD-L1 was highly expressed in EGFR wildtype NSCLC [35, 36]; meanwhile, a negative or no correlation between EGFR mutation and PD-L1 expression was also concluded [37‐39]. For this study, almost all cases with EGFR mutation were lung ACs, and no obvious difference of PD-L1 expression in EGFR wildtype and EGER mutant tumors was found in overall NSCLC and AC groups, respectively.

Furthermore, although the results of the present study demonstrated that PD-L1 expression of NSCLC was related with SUVmax, TLG, man, smoking, and central location, multivariate analysis displayed that only SUVmax was identified as an independent predictor of PD-L1 positivity in NSCLC, which was in agreement with the previous studies [18, 20]. It may be due to that SUVmax represents a maximally active portion of the tumor, and preferably predicts the tumor aggression. Moreover, similar to the previous reports [19, 21], SUVmax was also an independent and unique predictor of PD-L1 positive expression in AC and SCC subgroups, respectively. Subsequently, the ROC curves and the areas under the curve in this study suggested that SUVmax could predict PD-L1 expression in overall NSCLC, AC, and SCC with high sensitivity, specificity, and accuracy, respectively, which was also in line with the previous reports [18, 20, 21]. Therefore, the SUVmax of FDG PET/CT could be helpful for exploring the existence of immune checkpoints.

Clearly, there were also several limitations in this study. Firstly, based on immunohistochemical requirements of tumor specimens for detecting PD-L1 expression, the patients included in the present study were mainly at stage I–III; therefore, cases at stage IV were quite limited, which was also the limitation of the previous studies. Secondly, due to the invasion of biopsy, the metastatic lesions were usually not biopsied again in routine clinical practice once PD-L1 expression of the primary lesion confirmed. For this study, PD-L1 expression was confirmed by the biopsy of the primary lesion. Thus, the relationship between SUVmax and PD-L1 of metastatic lesions was not analyzed. Although PD-L1 expression may exist a difference in primary and metastatic tumors, Kim et al. [40] demonstrated that the concordance of PD-L1 expression between primary and metastatic pulmonary adenocarcinomas is high when using cutoff values of 1% and 50%. Thirdly, PD-L1 expression was regarded as an indicator of poor prognosis of lung cancer [41‐43]. Meanwhile, SUVmax, MTV, and TLG were also reported to have the prognostic values in NSCLC [44, 45]. However, the correlations of PD-L1 expression and PET/CT metabolic parameters with the prognosis were not analyzed, which will be carried out in the future study. Finally, although exploring the existence of PD-L1 could provide strategies to choose immunotherapy for NSCLC, responses to PD-1/PD-L1 antibodies were also noted in NSCLC with low or absent PD-L1 expression. However, due to the small population choosing immunotherapy in this study, the correlations of PET/CT metabolic parameters and immunotherapy were not investigated. Therefore, further large-scale, prospective studies of the role of FDG PET/CT in immunotherapy are warranted.

PD-L1 expression of NSCLC was significantly related with SUVmax, TLG, man, smoking, and central location. No obvious difference in MTV, age, serum tumor markers, tumor size, TNM stage, and genetic mutation was observed between PD-L1 positive and negative groups. Only SUVmax of FDG PET/CT was an independent predictor of PD-L1 expression, which could help to explore the existence of immune checkpoints.