Pattern of F-18 FDG Uptake in Colon Cancer after Bacterial Cancer Therapy Using Engineered Salmonella Typhimurium: A Preliminary In Vivo Study

Purpose Bacterial cancer therapy (BCT) research using engineered Salmonella typhimurium has increased in recent years. 2-Deoxy-2[18F] fluoro-D-glucose positron emission tomography (FDG PET) is widely used in cancer patients to detect cancer, monitor treatment responses, and predict prognoses. The aim of this pilot study was to investigate FDG uptake patterns in a mouse tumor model after BCT. Procedures. BCT was performed via the intravenous injection of attenuated S. typhimurium (SLΔppGpp/lux) into female mice bearing a tumor (derived from CT26 murine colon cancer cells) in the right thigh. 18F-FDG PET images acquired before BCT and at different time points after BCT. In vivo bioluminescence imaging confirmed bacterial presence in the tumor. The tumor volume, standardized uptake value (SUV) of FDG (SUVmax and SUVmean), early SUV reduction%, and normalized tumor volume change were analyzed. Results Early after BCT (1 or 2 days post-injection (dpi)), FDG tumor uptake decreased in 10 out of 11 mice and then increased at later stages. FDG uptake before BCT was correlated with normalized tumor volume change after BCT. Early FDG reduction% after BCT was correlated with normalized volume change after BCT. Conclusions Early after BCT, FDG tumor uptake decreased and then increased at later stages. The higher the FDG tumor uptake before BCT, the better the BCT response. FDG uptake patterns were related to tumor volume change after BCT. Therefore, FDG uptake was a good candidate for evaluating BCT.

A common bacteria strain for BCT is attenuated Salmonella typhimurium which is defective in ppGpp synthesis (ΔppGpp S. typhimurium) [20]. The bacteria displays a 100,000-1,000,000fold increased median lethal dose [21]. Zheng et al. reported that ΔppGpp S. typhimurium was more than 10,000-fold higher in tumor tissue when compared with other organs [8]. Critically, bacteria injected during BCT were rapidly cleared from internal organs [5], while after tumor targeting, BCT bacteria were not released from tumors and did not re-enter the blood circulation [22].
Our previous S. typhimurium in vivo research showed that tumor growth after BCT exhibited two phases: tumor growth suppression for 1-10 days, but regrowth thereafter [23]. The trajectory of individual tumor growth after BCT often shows significant variations in treatment responses [24]. However, no modalities are currently available to predict the tumor suppression effects of BCT; therefore, in vivo imaging techniques are required to monitor and predict BCT effects.
The glucose analog, 2-deoxy-2[ 18 F]fluoro-D-glucose (FDG) is widely used as a radiotracer for positron emission tomography (PET) and typically shows high FDG uptake in malignant tumors due to increased anaerobic glycolysis [25]. In patients with various cancers, FDG PET/computed tomography (PET/ CT) strategies are used for tumor detection, monitoring treatment responses, and prognosis predictions. FDG also accumulates in activated immune cells [26], whereas BCT mechanisms trigger immune responses [7,23,27]. Therefore, based on these uptake modalities, FDG appears to be a good candidate for evaluating BCT. To our knowledge, no previous reports have analyzed FDG uptake changes in tumors after BCT.
We therefore investigated FDG uptake patterns in tumors in a murine tumor model using attenuated S. typhimurium during and after BCT.

Study
Overview. This study was performed using two separate trials: experiment A (n = 6 mice) and B (n = 5 mice). We used this approach to avoid overburdening individual animals with different preparation procedures, such as fasting and general anesthesia ( Figure 1). After generating colon cancer animal model (murine CT26 colon adenocarcinoma cell line and BALB/c female mice; Section 2.2), BCT was initiated using a single intravenous injection (Section 2.3). PET/CT scans were acquired before BCT and serially during follow-up.
In trial B, PET/CT scans were acquired at 0 dpi, 1 dpi, 4 dpi, 7 dpi, and 15 dpi (b1-b5 mice). The tumor size was measured at 0 dpi, 3 dpi, 6 dpi, 9 dpi, 12 dpi, 16 dpi, and 19 dpi in trial A and at 0 dpi, 3 dpi, 6 dpi, 9 dpi, 12 dpi, 15 dpi, and 17 dpi in trial B.  Figure 1: Study overview. This pilot study was performed using two separate trials: A and B. A tumor was generated in the right thigh of each mouse (BALB/c female mice, 6-weeks-old) using the murine CT26 colon adenocarcinoma cells. After tumor volume reached approximately 180 mm 3 , BCT was performed by intravenous injection of bacterial cells via the tail vein (bacterial dose = 4:5 × 10 7 CFU/ mouse in 100 μL PBS). Two trials were designed to avoid animal preparation issues such as fasting and general anesthesia. In trial A (n = 6), FDG PET/CT images were acquired at 0 dpi, 2 dpi, 10 dpi, and 16 dpi (a). In trial B (n = 5), FDG PET/CT images were acquired at 0 dpi, 1 dpi, 4 dpi, 7 dpi, and 15 dpi (b). In vivo BLi confirming the presence of tumor-targeting bacteria was performed at 2 dpi, immediately after PET/CT in trial A, and immediately after each PET/CT in trial B. Abbreviations: FDG, 2-deoxy-2[ 18 F]fluoro-Dglucose; dpi, days post-injection; SUVmax, maximum standardized uptake value; SUVmean, mean standardized uptake value; PBS, phosphate-buffered saline; CFU, colony-forming units; BLi, bioluminescence imaging.  [30,31]. After manually drawing a region-of-interest (ROI) in the tumor at the highest uptake point, maximum standardized uptake (SUVmax) and mean standardized uptake values (SUVmean) were obtained using PMOD.

The
2.5. In Vivo Bioluminescence (BLi) Imaging. In vivo BLi confirmed successful injection and SLΔppGpp/lux accumulation in tumors. Images were acquired after PET/CT scanning on the same day. To capture images, an IVIS-100 imaging system equipped with a charged coupled device camera (Caliper Life Sciences, Waltham, MA, USA) was used. After anesthetization with 2.5% isoflurane in air, mice were placed in the light-tight chamber of the IVIS-100 system. Living image software v.2.25 (Caliper Life Sciences) was used for image acquisition and processing. In trial A, in vivo BLi was acquired at 2 dpi. In trial B, in vivo BLi was performed on the same day as PET scanning and in vivo BLi counts were also acquired.
2.6. Analysis of Variables. SUVmax, SUVmean, and tumor volume values at each time point were used for analyses.
The following formula is used: A circular ROI was drawn over the FDG uptake area in the tumor, after which SUVmax and SUVmean values were obtained.

Experiment for Bacterial Load/Colonization Efficiency.
For the evaluation of efficiency of bacterial load, an additional experiment was done after trials A and B. CT26 xenografts were generated in 9 mice. SLΔppGpp/lux was injected into tumor-bearing mice in the same way. In vivo BLi was done at 1 dpi in mouse 1~3, at 3 dpi in mouse 4~6, and at 5 dpi in mouse 7~9. After in vivo BLi, the blood, lung, liver, spleen, and tumor of each mice were extracted and imaged by ex vivo BLi.

Statistical
Analysis. Spearman's rank correlation coefficient (rho) was used to analyze correlations between nonparametric variables. All statistical analyses were performed using Med-calc® v.18.5. Statistical significance was determined at p < 0:05.

Results
3.1. General Features before BCT. Between trials A and B, we observed no significant differences in tumor volume (0 dpi). Among the 11 mice, two died during the study: A5 at 2 dpi and A3 at 16 dpi (size measurements were performed until 12 dpi). Tumor volume at 0 dpi was not correlated with SUVmax (p > 0:5) or SUVmean (p > 0:5) at 0 dpi.

Tumor
Growth and BLi after BCT. In vivo BLi confirmed successful SLΔppGpp/lux accumulation in all mice tumors. After BCT, tumor growth was suppressed until 6-9 dpi in most mice, with growth then recommencing (Figures 2  and 3). However, in a1 and a3 mice, tumors were not suppressed and continued to grow (Figure 2(a)).

FDG Uptake Patterns in Tumors after BCT.
FDG uptake in tumors decreased in the early days after BCT (1 or 2 dpi) in 10 of 11 mice, but it increased in later days. In only one mouse (a3), FDG uptake did not decrease in the early days after BCT (Figure 3(c)). Between trials A and B, we observed no significant differences in hepatic FDG uptake and lung FDG uptake ( Figure S2, ESM).

Correlations between FDG Uptake after BCT and Treatment Responses.
After BCT, FDG uptake in tumors decreased in the early days (1 or 2 dpi). We calculated this using early SUVmax reduction% and early SUVmean reduc-tion%. We also observed a significant correlation between normalized tumor volume change and early SUV reduction % ( Figure 5). At the early response time point (6 dpi), both early SUVmax reduction% and early SUVmean reduction% were correlated with normalized tumor volume change (rho = 0:685 and p = 0:0289 for SUVmax reduction% and rho = 0:891 and p = 0:0005 for SUVmean reduction%;  5(d)). Taken together, these data indicated that a reduction in FDG uptake in the early days after BCT predicted BCT outcomes.

Discussion
To our knowledge, this is the first pilot study to analyze FDG uptake changes in tumors after BCT. We observed several unique FDG uptake features: (1) Tumor FDG uptake decreased in the first 2 days after BCT and then increased, before tumor graft re-growth; (2) FDG uptake before BCT was correlated with normalized tumor volume change after BCT; and (3) early FDG reduction% after BCT was correlated with normalized tumor volume change after BCT.
The mechanisms underlying tumor-targeting bacteria may involve the chemotactic system [32][33][34], inflammatory cytokine-mediated dilation of tumor blood vessels [35,36], and immune-privileged tumor environments [7,37,38]. However, the mechanism underpinning the therapeutic effects of BCT using Salmonella species is unclear. Salmonella species are believed to kill tumor cells by apoptosis and/or autophagy via alterations in host antitumor immune responses or nutrient deprivation [1]. Salmonella also activates the inflammasome pathway via damage signal release from cancer cells and macrophages [39]. These effects may contribute to the BCT therapeutic effects of Salmonella.
FDG is not a specific radiotracer for apoptosis. However, BCT mechanisms are not simply limited to apoptosis. A recent SUVmax, maximum standardized uptake value; SUVmean, mean standardized uptake value; early SUVmax reduction%, 100 × ðSUV max at 0 dpi − SUV max at 1 or 2 dpiÞ/SUVmax at 0 dpi; early SUVmean reduction%, 100 × ðSUV mean at 0 dpi − SUV max at 1 or 2 dpiÞ/SUVmean at 0 dpi. 8 Molecular Imaging study investigating apoptosis imaging tracers reported that FDG was more reliable and sensitive for evaluating therapeutic effects [40]. The authors of this study reported that tumor apoptosis, induced by an antiangiogenic agent, was more sensitively and reliably monitored by FDG when compared with Annexin V-based apoptosis imaging [40]. Furthermore, FDG uptake in malignant tumors was influenced not only by cancer cells but also cells in the tumor microenvironment [41][42][43]. Activated neutrophils increased GLUT type 3 and 4 expression resulting in increased glucose uptake [42], and tumor necrosis factor-α secreted by macrophages increased FDG uptake in tumor cells [43]. Lymphocytes also increased FDG uptake in tumors based on their numbers and activation status [41]. Therefore, elevated FDG accumulation in tumors was reflected by the high numbers and activities of both cancer and immune cells. Based on these proposed uptake mechanisms, FDG is a good candidate for evaluating BCT. We identified an early (within 1 or 2 days) decrease in FDG uptake in tumors after BCT. This reflected Salmonella actions in the early period; bacterial accumulation in tumors was confirmed by in vivo BLi at 1 dpi. Ganai et al. reported that S. typhimurium accumulated in tumors 3 hours after systemic injection [32], and at up to 48 hours later, tumor apoptosis had increased and viable tissue decreased [32]. These data were consistent with our findings of an early decrease in FDG uptake in tumors after BCT.
Previous studies also reported that high tumor FDG uptake was associated with poor survival in patients with lung cancer, breast cancer, colon cancer, or lymphoma [44]. In addition, tumor FDG uptake correlated with the levels of several prognostic factors such as p53, Ki67, GLUT1, and hexokinase in patients with colon cancer [45]. In contrast, in this study, tumors with high FDG uptake before BCT showed better results after BCT. We observed that FDG uptake before BCT correlated with normalized tumor volume changes after BCT. High FDG uptake before BCT showed better results after BCT and suggested BCT was more effective in tumors with increased glucose metabolism. This discrepancy was understandable due to the important role of immune cells in the tumor microenvironment in BCT when compared with conventional treatments, such as chemotherapy and radiotherapy. In BCT, activated immune cells are mainly involved in tumoricidal mechanisms [1,39]. A high FDG uptake represents high numbers of activated immune cells [46,47]. In previous studies, we also reported that BCT outcomes correlated well with immune cell infiltration and activation in the tumor milieu [8,23]. Our findings also indicated different treatment mechanisms between conventional cancer therapies and BCT. Additionally, when we analyzed in vivo BLi optical signals in trial B, they were positively correlated with FDG uptake in the tumor before BCT, suggesting that tumor targeting by S. typhimurium was better in hypermetabolic tumors.
We observed that early SUV reduction% correlated with normalized tumor volume change after BCT. This meant that a higher reduction in FDG uptake in the early days after BCT predicted a smaller tumor mass after BCT. We hypothesized that early bacterial reactions could determine the final effects of BCT and this information could be used to predict BCT effects. This pilot study had some limitations. The number of study animals was small. Further imaging studies are underway to compare the characteristics of different radiotracers such as FDG and fluorodeoxysorbitol in BCT. Also, we did not perform in vitro histological and genomic analyses before and after BCT. However, our study was exclusively focused on the analysis of FDG uptake patterns in BCT. Further research investigating in vitro molecular genomic changes following BCT is warranted.

Conclusions
This was the first study investigating FDG uptake patterns in BCT using attenuated S. typhimurium. Early after BCT, FDG tumor uptake decreased and then increased at later stages. The higher the FDG uptake before BCT, the better the BCT response, and FDG uptake patterns were related to tumor volume change after BCT. Therefore, FDG uptake appears to be a good candidate for BCT evaluating BCT.

Data Availability
The data is available on request.

Ethical Approval
All applicable institutional and/or national guidelines for the care and use of animals were followed. All animal experiments were conformed to the Chonnam National University Animal Research Committee protocols. Figure S1: accumulation of engineered Salmonella typhimurium in a CT26 colon cancer model. Tumor targeting of engineered S. typhimurium (SLppGpp-lux) in a CT26 colon cancer model was confirmed by in vivo BLi from 1 dpi after BCT. Data are showing bacterial load/colonization efficiency at the tumor after BCT in a CT26 colon cancer model. Figure S2: comparison of tumor volume and liver and lung FDG uptake before BCT in a CT26 colon cancer model between trial A and B. Figure S3: correlation between pretreatment tumor volume and tumor FDG uptake before treatment and treatment results in a CT26 colon cancer model. Figure S4: correlation between pretreatment FDG uptake and tumor volume after BCT in a CT26 colon cancer model. Figure S5: correlation between early SUV reduction% and tumor volume after BCT in a CT26 colon cancer model. 9 Molecular Imaging Figure S6: changes in FDG uptake in liver and lung before and after BCT in a CT26 colon cancer model. (Supplementary  Materials)