Prognostic Value of Dual-Time-Point [18F]FDG PET/CT for Predicting Distant Metastasis after Treatment in Patients with Non-Small Cell Lung Cancer

This study aimed to evaluate the prognostic significance of 2-Deoxy-2-[18F]fluoro-D-glucose ([18F]FDG) uptake in the bone marrow (BM) and primary tumors on dual-time-point (DTP) PET/CT for predicting progression-free survival (PFS) and distant metastasis-free survival (DMFS) in patients with non-small cell lung cancer (NSCLC). We retrospectively analyzed DTP [18F]FDG PET/CT images from 211 patients with NSCLC. The maximum standardized uptake value (SUV) of primary lung cancer and mean [18F]FDG uptake of the BM (BM SUV) were measured from early and delayed PET/CT images, and the percent changes in these parameters (∆maximum SUV and ∆BM SUV) were calculated. On multivariate survival analysis, the maximum SUV and BM SUV on both early and delayed PET/CT scans were significantly associated with PFS, while the ∆maximum SUV and ∆BM SUV failed to show statistical significance. For DMFS, the ∆maximum SUV and ∆BM SUV were independent predictors along with the TNM stage. Distant progression was observed only in 1.3% of patients with low ∆maximum SUV and ∆BM SUV, whereas 28.2% of patients with high ∆maximum SUV and ∆BM SUV experienced distant progression. The ∆maximum SUV and ∆BM SUV on DTP [18F]FDG PET/CT were significant independent predictors for DMFS in patients with NSCLC.


Introduction
Non-small cell lung cancer (NSCLC) is the most common histopathological category of lung cancer and is notable for its poor clinical outcomes, with a 5-year overall survival rate between 10-20% [1]. With advancements in treatment modalities and strategies for NSCLC, a number of studies have investigated prognostic factors that could predict the risk of cancer progression and guide clinical decisions for managing patients with NSCLC [2]. Recently, several studies have shown that the inflammatory response of the host by the activation of immune cells plays an essential role in the development, growth, and metastasis of NSCLC cells [3][4][5]. In clinical observational studies, diverse serum inflammatory markers such as Creactive protein (CRP), neutrophil-to-lymphocyte ratio (NLR), and platelet-to-lymphocyte ratio (PLR) turned out to be significantly related with clinical outcomes in patients with NSCLC, irrespective of the tumor stage and treatment modalities [6][7][8]. These results Retrospective review of the medical records was performed for 493 subjects who had DTP [ 18 F]FDG PET/CT for the characterization of pulmonary nodules and/or mass lesions between March 2014 and August 2020 at CHA Bundang Medical Center. Of them, we finally recruited 211 patients (1) who were histopathologically diagnosed with NSCLC, (2) who showed no distant metastasis on staging examinations, and (3) who underwent surgical resection, radiotherapy, and/or chemotherapy for NSCLC after staging work-up. Patients (1) who had a history of another malignant disease, (2) who lacked blood test results including CRP and blood cell counts within a month before or after DTP [ 18 F]FDG PET/CT, (3) who received supportive care without any curative or palliative treatment, and (4) who were lost to follow-up within 12 months after the treatment without recurrence were excluded from the study. Before the initial treatment, staging work-up comprising blood tests, bronchoscopy, contrast-enhanced chest CT, brain magnetic resonance imaging (MRI), and DTP [ 18 F]FDG PET/CT was performed for the enrolled patients. The patients were treated with surgery, chemoradiotherapy, radiotherapy, or chemotherapy based on their clinical stage and condition. The mean interval between DTP PET/CT and the initial treatment was 11 ± 9 days. After treatment, the patients were followed-up with imaging studies at regular intervals of 3-6 months. Disease progression of NSCLC was diagnosed based on imaging examinations and/or histopathological confirmation. The patients with disease progression were categorized into two groups: those with locoregional disease progression and no distant metastatic lesions (locoregional progression group), and those with distant metastatic lesions irrespective of locoregional disease progression (distant progression group).

Hematologic Features
Blood laboratory data were obtained from blood tests at the staging work-up within a month before or after DTP [ 18  F]FDG injection from the skull base to the mid-thigh and the delayed PET/CT scan was performed approximately 120 min after the injection for the chest region. A CT scan was performed at 100 kVp and 40 mA without contrast injection followed by a PET scan for 90 s in each bed position using the three-dimensional acquisition mode. PET images were reconstructed using an iterative ordered subset expectation maximization algorithm with attenuation correction.

PET/CT Image Analysis
Two nuclear medicine physicians, who were unaware of the clinicohistopathological characteristics and survival results of the patients, retrospectively evaluated all DTP [ 18 F]FDG PET/CT images using OsiriX MD 10.11 software (Pixmeo, Geneva, Switzerland). For each patient, the maximum standardized uptake value (SUV) of the lung cancer and mean [ 18 F]FDG uptake of the BM (BM SUV) were measured on both early and delayed PET/CT images. The volume of interest (VOI) was manually drawn on the primary tumor, and the maximum SUV of the tumor was measured. Subsequently, a spheroidshaped VOI was placed over the vertebral body of five thoracic spines (T8-T12 spines), and an isocontour using a cut-off SUV of 75% of the maximum SUV of the VOI was automatically drawn within the VOI for each thoracic spine [18,19]. The mean SUV of voxels within the isocontour in the VOIs of the five selected thoracic spines was defined as the BM SUV. In patients who showed postoperative changes from previous spinal surgeries, compression fractures, and severe osteoarthritic changes in T8-T12 spines, other thoracic spines with no abnormal findings were selected for the measurement of the BM SUV [18,19]. Using the maximum SUV of the primary lung cancer and BM SUV on the early and delayed PET images, the percent change in both PET parameters (∆maximum SUV and ∆BM SUV) between the early and delayed PET scans was calculated as follows: (∆PET parameter) = ((parameter on delayed PET scan)-(parameter on early PET scan))/(parameter on early PET scan) × 100.

Statistical Analysis
We calculated Spearman rank correlation coefficients to evaluate the relationships between the BM parameters on DTP PET/CT images and serum inflammatory markers after assessing the normality of distribution. The Kruskal-Wallis test and post hoc comparisons with Dunn's test were performed to assess the differences in DTP PET/CT parameters between the patient groups with no progression, locoregional progression, and distant progression. The prognostic significance of DTP [ 18 F]FDG PET/CT parameters and clinicohistopathological factors for predicting PFS and DMFS was assessed using univariate and multivariate Cox proportional hazards regression analyses. For multivariate analysis, only variables that revealed a statistical significance on univariate survival analysis were included after adjusting for age and sex. Survival time was defined as the time from the day of initial treatment to the day of the detection of disease progression (distant progression for DMFS) or death from any cause. Patients with no disease progression were censored on the day of the last clinical follow-up. The optimal cut-off values of the PET/CT parameters were chosen by receiver operating characteristic (ROC) curve analysis, and the Kaplan-Meier method was used to estimate the survival curves of the PET/CT parameters to calculate the cumulative DMFS. The patients were classified into four subgroups based on the cut-off values of ∆maximum SUV and ∆BM SUV, and Fisher's exact test was performed to assess the differences in distant metastasis rates between the patient subgroups. All data were statistically analyzed using MedCalc Statistical Software version 20.014 (MedCalc Software Ltd., Ostend, Belgium), and p-values of <0.05 were regarded statistically significant.

Clinical Characteristics
The clinicohistopathological characteristics and DTP [ 18 F]FDG PET/CT parameters of the 211 enrolled patients with NSCLC are summarized in Table 1. In pairwise comparisons, the maximum SUV of the primary lung tumors and BM SUV on delayed PET/CT scans were significantly higher than those on early PET/CT scans (p < 0.001 for both). Of all patients, 15 (7.  BM SUV, mean standardized uptake value of bone marrow; CRP, C-reactive protein; CT, computed tomography; NLR, neutrophil-to-lymphocyte ratio; PET, positron emission tomography; PLR, platelet-to-lymphocyte ratio; SUV, standardized uptake value; TNM, tumor, nodes, and metastases; WBC, white blood cell. Chemotherapy alone 17 (8.1%) Radiotherapy alone 12 (5.7%) BM SUV, mean standardized uptake value of bone marrow; CRP, C-reactive protein; CT, computed tomography; NLR, neutrophil-to-lymphocyte ratio; PET, positron emission tomography; PLR, platelet-to-lymphocyte ratio; SUV, standardized uptake value; TNM, tumor, nodes, and metastases; WBC, white blood cell.

Correlation Analysis
The relationships between three BM parameters (BM SUV on early and delayed PET/CT scans and ∆BM SUV) and four serum inflammatory markers (CRP, WBC counts, NLR, and PLR) were evaluated ( Table 2). The BM SUV on early and delayed PET/CT scans showed significant positive correlations with the CRP, WBC count, and NLR, whereas no significant correlation was observed with the PLR (p > 0.05). In contrast, the ∆BM SUV did

Correlation Analysis
The relationships between three BM parameters (BM SUV on early and delayed PET/CT scans and ∆BM SUV) and four serum inflammatory markers (CRP, WBC counts, NLR, and PLR) were evaluated ( Table 2). The BM SUV on early and delayed PET/CT scans showed significant positive correlations with the CRP, WBC count, and NLR, whereas no significant correlation was observed with the PLR (p > 0.05). In contrast, the ∆BM SUV did not show significant correlations with any of the four serum inflammatory markers (p > 0.05). BM SUV, mean standardized uptake value of bone marrow; CRP, C-reactive protein; CT, computed tomography NLR, neutrophil-to-lymphocyte ratio; PET, positron emission tomography; PLR, platelet-to-lymphocyte ratio; SUV, standardized uptake value; WBC, white blood cellThe relationship between the disease progression pattern and DTP PET/CT parameters was also assessed. The results of the Kruskal-Wallis test demonstrated that all DTP PET/CT parameters showed significant differences among patients with no progression, locoregional progression, and distant progression (p < 0.05; Table 3). Post hoc analysis showed that patients with locoregional and distant progression had significantly higher values of all PET/CT parameters than those with no progression (p < 0.05).
On comparison between patients with locoregional and distant progression, there were no significant differences in the maximum SUV of the primary tumor and the BM SUV on early and delayed PET/CT (p > 0.10). Meanwhile, patients with distant progression tended to show higher ∆maximum SUV and ∆BM SUV values than those with locoregional progression with a borderline statistical significance (p = 0.092 for ∆maximum SUV and p = 0.064 for ∆BM SUV; Figure 2).

Survival Analysis
The prognostic significance of DTP [ 18 F]FDG PET/CT parameters and clinicohisto-

Survival Analysis
The prognostic significance of DTP [ 18 F]FDG PET/CT parameters and clinicohistopathological factors for predicting PFS and DMFS on univariate and multivariate analyses is shown in Tables 4 and 5. On univariate analysis, all six DTP PET/CT parameters showed significance in predicting PFS, along with age, sex, TNM stage, treatment, CRP level, WBC count, NLR, and PLR (p < 0.05; Table 4). For DMFS, the BM SUV on delayed PET/CT, ∆maximum SUV, and ∆BM SUV were significant prognostic factors, along with the TNM stage and NLR (p < 0.05; Table 4). Among the variables, those that showed statistical significance on univariate analysis were included on multivariate analysis. As the maximum SUV and BM SUV on early and delayed PET/CT scans were significantly correlated (p < 0.001, r = 0.905 for maximum SUV and p < 0.001, r = 0.847 for BM SUV), two separate multivariate models were used for predicting PFS: one with early PET/CT parameters and another with delayed PET/CT parameters. On multivariate analysis, the TNM stage, treatment, and maximum SUV of the primary tumor and BM SUV on both early and delayed PET/CT showed a significant association with PFS (p < 0.05), whereas both ∆maximum SUV and ∆BM SUV did not show prognostic significance (p > 0.05; Table 5). For DMFS, only the ∆maximum SUV (p = 0.008) and ∆BM SUV (p = 0.009) were independent predictors, along with the TNM stage (Table 5). An increase in the ∆maximum SUV and ∆BM SUV was associated with an increased risk of distant metastasis after treatment.
For Kaplan-Meier analysis, the ∆maximum SUV and ∆BM SUV were categorized into two groups according to the specific cut-off values (30.00 for ∆maximum SUV and 27.50 for ∆BM SUV). The results of the Kaplan-Meier analysis demonstrated that patients with a high ∆maximum SUV and ∆BM SUV had significantly worse DMFS than those with low values (p = 0.002 for ∆maximum SUV and p < 0.001 for ∆BM SUV; Figure 3).     A combination of the ∆maximum SUV and ∆BM SUV could further enhance the predictive value for the risk of distant metastasis. We categorized the patients into four subgroups according to the cut-off values of the ∆maximum SUV and ∆BM SUV and compared the distant metastasis rates between those four groups ( Table 6). The patient subgroup with a high ∆maximum SUV and ∆BM SUV showed the highest distant metastasis rate (28.2%) after treatment among the four subgroups. This rate was significantly higher than that in the other three patient subgroups (subgroup with low ∆maximum SUV and ∆BM SUV: 1.3%, p < 0.001; subgroup with high ∆maximum SUV and low ∆BM SUV; 7.0%, p = 0.008; and subgroup with low ∆maximum SUV and high ∆BM SUV: 8.6%, p = 0.040).

Discussion
In the present study, we investigated the prognostic significance of the BM SUV as well as the maximum SUV of NSCLC on DTP [ 18 F]FDG PET/CT for predicting PFS and DMFS. For PFS, the maximum SUV and BM SUV on both early and delayed PET/CT were independent predictors, whereas ∆maximum SUV and ∆BM SUV did not show a significant association. In contrast, for DMFS, only the ∆maximum SUV and ∆BM SUV were independent prognostic factors, and the maximum SUV and BM SUV did not have any significant prognostic value. In the literature, this is the first study to demonstrate a significant relationship between changes in the maximum SUV of the primary tumor and BM SUV on DTP PET/CT with the risk of distant metastasis after treatment in patients with solid cancers.
In previous studies, [ 18 F]FDG uptake in the BM on PET/CT was found to not only correlate with the extent of systemic inflammatory response but also reflect the immune response to cancer [11,[20][21][22]. In the correlation analysis, [ 18 F]FDG uptake in the BM showed significant positive correlations with diverse serum inflammatory markers including NLR, CRP, WBC count, and PLR [11,14,15,23], which was also observed in our study. Furthermore, on survival analysis, [ 18 F]FDG uptake in the BM was an independent prognostic factor for predicting PFS and overall survival in patients with head and neck, breast, lung, pancreatic, colorectal, and gynecologic cancers, consistently demonstrating worse survival in patients with a high [ 18 F]FDG uptake in the BM [14,15,[18][19][20]22,24]. For NSCLC, [ 18 F]FDG uptake in the BM showed significant positive associations with clinical outcomes in patient cohorts treated with curative surgery, definite chemoradiotherapy, and chemotherapy [11,12,21]. In accordance with previous studies, our study also revealed that the BM SUVs on DTP PET/CT scans were significant prognostic factors for predicting PFS in multivariate survival analysis. In a previous study that enrolled patients with gynecologic cancers, those with a high [ 18 F]FDG uptake in the BM demonstrated increased numbers of immunosuppressive immune cells, specifically myeloid-derived suppressor cells, and decreased numbers of antitumoral immune cells, such as cytotoxic T cells, in the peripheral blood and tumors [22]. Considering that the BM can regulate the balance between protumoral and antitumoral immunity, it could be assumed that increased [ 18 F]FDG uptake of the BM in patients with cancer reflects enhanced immunosuppressive immune cell activities in the BM [25,26].
In addition to the significant association between PFS and BM SUV, the results of this study demonstrated that ∆BM SUV was a significant prognostic indicator for predicting DMFS. To date, only one study has investigated the prognostic value of changes in ∆BM SUV on DTP PET/CT [27]. This study enrolled 70 patients with diffuse large B-cell lymphoma and showed that patients with a ∆BM SUV of >45 had shorter overall survival than others [27]. Although the underlying mechanism for the relationship between ∆BM SUV and survival is yet to be elucidated, two possible explanations derived from the host and tumor factors could be suggested. Increased ∆BM SUV might represent the host BM conditions with highly activated immune cells, especially myeloid-derived suppressor cells and tumor-associated macrophages that promote epithelial-mesenchymal transition in NSCLC cells [28,29]. The aforementioned previous study also suggested that the increased ∆BM SUV could be attributable to the activation of the metabolic status of BM cells, which is induced by proinflammatory cytokines [27]. Another possible explanation is the presence of hidden metastatic cancer cells in the BM. Although patients who showed distant metastatic lesions on staging imaging examinations were excluded from the study, hidden microscopic metastasis might have been present in some of the enrolled patients, which would have affected the increase in ∆BM SUV and the risk of distant metastasis [30]. However, further studies are required to clarify the main mechanisms underlying this relationship.
Along with the ∆BM SUV, the ∆maximum SUV of primary lung cancer is another imaging parameter of DTP PET/CT that showed a significant association with DMFS on multivariate survival analysis. In previous studies, the maximum SUV and ∆maximum SUV of tumors on DTP [ 18 F]FDG PET/CT were found to be related to different aspects of cancer cell metabolism [31,32]. The maximum SUV of tumors on early PET/CT scans was mainly related to glucose transporter-1 expression in tumor cells, whereas the ∆maximum SUV had a close positive relationship with the expression ratio of hexokinase to glucose-6-phosphatase [31,32]. As the degree of hexokinase expression is related to the biological aggressiveness and proliferation rate of cancer cells, ∆maximum SUV is proposed as a promising imaging parameter that can reflect the malignant potential of cancer lesions [16,[31][32][33]. However, clinical studies on patients with NSCLC regarding the prognostic value of the ∆maximum SUV have demonstrated inconsistent results. The ∆maximum SUV was shown to be an independent prognostic factor for predicting PFS and recurrencefree survival in several studies; however, other studies revealed that the ∆maximum SUV did not have any prognostic value for survival [13,16,33,34]. Another previous study that enrolled 57 patients with stage I NSCLC who had received stereotactic body radiation therapy demonstrated that the ∆maximum SUV was significantly associated with the risk of distant metastasis after radiotherapy, which is similar to the results of our study [35]. Furthermore, the study also revealed that the ∆maximum SUV tended to predict a lower risk of local recurrence and regional lymph node metastasis in NSCLC [35]. Considering these results, the ∆maximum SUV might have an inferior prognostic value compared with maximum SUV for predicting the risk of overall disease progression; however, it could be used as a potential indicator of distant progression of NSCLC.
There is now ample evidence to suggest that the response of host immune cells to cancer cells, as well as the intrinsic features of cancer cells, is crucial for cancer progression and metastasis [36]. Based on this theory, recent studies have attempted to integrate imaging findings of primary tumors with imaging parameters that reflect the host's immune response for a more precise prediction of clinical outcomes in patients with malignant diseases [14,20,37,38]. In previous studies of breast and pancreatic cancers, the risk of distant recurrence and death could be further stratified by merging BM findings with primary tumor imaging features [14,20]. Another study on patients with NSCLC also demonstrated that a model including imaging features of both primary tumors and BM more accurately predicted disease-free survival than clinical features alone [37]. Similarly, we could further stratify the risk of distant metastasis after treatment by combining the ∆maximum SUV of the primary tumor and the ∆BM SUV in patients with NSCLC. Our results showed that the distant metastasis rate in the patient subgroup with a high ∆maximum SUV and ∆BM SUV was up to 28.2%, which was significantly higher than that in all other patient subgroups. Although patients who experience distant progression have worse survival than those with locoregional progression, recent advances in NSCLC treatment are expected to lead to improved survival in patients with distant metastasis [39,40]. Therefore, more careful surveillance and intensive therapeutic intervention could be needed in patients with a high ∆maximum SUV and ∆BM SUV. Furthermore, when disease progression is suspected in these patients, whole-body imaging modalities such as [ 18 F]FDG PET/CT should be preferentially considered for the possibility of distant progression.
Our study has several limitations. First, as this was a retrospectively performed singlecenter study, which enrolled a limited number of patients, further studies are necessary to confirm our results. Second, TNM stage and treatment modalities of the enrolled patient population were heterogeneous, which might affect the results of the study. Third, because distant progression was diagnosed mainly by imaging studies, it is possible that the exact incidence of distant progression was underestimated. Fourth, previous studies have generally used 120 min after [ 18 F]FDG injection as the imaging time for delayed PET/CT scans [13,16,17,34,35]; however, there is still no consensus on the best imaging time for the delayed PET/CT scan [32,33]. Finally, further investigations are required to identify the biological mechanisms underlying our results.

Conclusions
The changes in the maximum SUV of the primary tumor and BM SUV were significantly associated with the risk of distant progression after treatment in patients with NSCLC. Patients with high ∆maximum SUV and ∆BM SUV had worse DMFS than those with low values. The ∆maximum SUV and ∆BM SUV on DTP [ 18 F]FDG PET/CT might provide prognostic information regarding distant progression in patients with NSCLC.

Institutional Review Board Statement:
The study was conducted in accordance with the Declaration of Helsinki, and approved by the Institutional Review Board of CHA university (Protocol code: CHAMC 2020-06-032, date of approval: 24 June 2020).
Informed Consent Statement: Patient consent was waived by the Institutional Review Board due to the retrospective design of the study.

Data Availability Statement:
The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.