In vivo imaging of invasive aspergillosis with 18F-fluorodeoxysorbitol positron emission tomography

Invasive aspergillosis is a critical complication in immunocompromised patients with hematologic malignancies or with viral pneumonia caused by influenza virus or SARS‑CoV‑2. Although early and accurate diagnosis of invasive aspergillosis can maximize clinical outcomes, current diagnostic methods are time-consuming and poorly sensitive. Here, we assess the ability of 2-deoxy-2-18F-fluorosorbitol (18F-FDS) positron emission tomography (PET) to specifically and noninvasively detect Aspergillus infections. We show that 18F-FDS PET can be used to visualize Aspergillus fumigatus infection of the lungs, brain, and muscles in mouse models. In particular, 18F-FDS can distinguish pulmonary aspergillosis from Staphylococcus aureus infection, both of which induce pulmonary infiltrates in immunocompromised patients. Thus, our results indicate that the combination of 18F-FDS PET and appropriate clinical information may be useful in the differential diagnosis and localization of invasive aspergillosis.

F ungal infections may be life-threatening in various types of patients, including patients with hematologic malignancies; those who have undergone solid organ or hematopoietic stem cell transplantation; those receiving chemotherapy, immunosuppressive therapy, or prolonged corticosteroid treatment; and those with prolonged stay in the intensive care unit (ICU) [1][2][3] . Aspergillus fumigatus (A. fumigatus) is the most frequent cause of invasive fungal disease in humans, usually affecting the lungs, sinuses, and brain 4 . Global models estimate that more than 3,000,000 individuals experience chronic pulmonary aspergillosis and over 250,000 experience invasive aspergillosis (IA) annually, with these infections associated with high morbidity and mortality rates 5 . IA is a frequent complication in immunocompromised patients such as those with neutropenia 2,6 , but the incidence of IA in non-neutropenic patients with underlying lung diseases (e.g., chronic obstructive pulmonary disease, asthma, lung cancer, and viral pneumonia) 7,8 has also increased. Several recent reports have described COVID-19-associated pulmonary aspergillosis (CAPA) [9][10][11][12] , raising concerns that CAPA could worsen outcomes of COVID-19 13 .
Early recognition of IA and immediate administration of proper antifungal agents are crucial for patient survival, but early recognition is usually difficult because the pace of progression leaves only a narrow window for diagnosis 14 . Moreover, invasive procedures for definitive diagnosis are not possible due to patient comorbidities, such as coagulopathy and respiratory compromise 14 . Although mycological culture of Aspergillus from infected tissues is the reference standard for IA diagnosis, this method is invasive, with low sensitivity and a long turnaround time 7,14 . The detection of circulating antigens such as galactomannan (GM) and β-D-glucan (BDG) is regarded as a rational first step for diagnosis of IA, but this indirect microbiologic test has low sensitivity and specificity 7,14 . Although computed tomography (CT) is frequently used to make decisions regarding the treatment of invasive pulmonary aspergillosis (IPA), radiographic findings in IPA are generally non-specific and differ by neutrophil count 15,16 . In patients with mixed infection such as CAPA, the lack of specific radiologic signs of IPA complicates its radiologic diagnosis 13 . Positron emission tomography (PET) using 2-deoxy-2-18 F-fluoro-D-glucose ( 18 F-FDG), a successful molecular imaging method in oncology, was unable to distinguish IPA from sterile inflammation, cancer, and bacterial infection in both preclinical and clinical studies [17][18][19] . Other molecular imaging tracers, such as 99m Tc-labeled phosphorodiamidate morpholino (MORF) oligomer probes 20 , 68 Ga-and 89 Zr-labeled siderophores [21][22][23][24] , and the 64 Cu-DOTA-labeled Aspergillus-specific mouse monoclonal antibody (mAb) mJF5 25 , have been developed, each tracer has limitations, including a lack of signal intensification at the site of infection and/or unknown toxicity in humans.
Because the sugar-free sweetener sorbitol is a metabolic substrate for Enterobacterales, the ability of 2-deoxy-2-18 F-fluorosorbitol ( 18 F-FDS) was assessed for its ability to image pathogenic E. coli 26 , with this method found to be selective for Gram-negative pathogenic bacteria 27,28 . Intriguingly, sorbitol can act as a sole carbon source supporting high fungal growth and sporulation 29 . The gene mpkC, which encodes a mitogenactivated protein kinases, was found to play a key role in controlling the growth of A. fumigatus in response to sorbitol as a carbon source 30 . These findings suggested that 18 F-FDS has potential as a suitable PET probe to image A. fumigatus infections in vivo.
Here, we evaluate the in vitro and in vivo properties of 18 F-FDS as an imaging agent for A. fumigatus infection in various mouse models. We show that 18 F-FDS selectively concentrates in regions of A. fumigatus infection and differentiates A. fumigatus infection from other infectious and non-infectious causes. Our results suggest that this radiotracer can be used as a PET molecular imaging agent to obtain high quality images in the diagnosis of A. fumigatus infection.

Results
18 F-FDS preparation and in vitro uptake studies. The identity and radiochemical purity of 18 F-FDS were evaluated by comparing the radio-TLC R f value of 18 F-FDS with that of cold 19 F-FDS 31,32 . The total time required to prepare 18 F-FDS was within 30 min, and its overall decay-corrected radiochemical yield was approximately 35-40%. 18 F-FDS had a specific activity greater than 13.8 ± 1.3 GBq/μmol and a radiochemical purity greater than 98%.
The in vitro 18 F-FDS uptake by A. fumigatus, Rhizopus arrhizus (R. arrhizus), and Candida albicans (C. albicans), clinically important fungal pathogens in immunocompromised patients, was assessed, with Escherichia coli (E. coli) and Staphylococcus aureus (S. aureus) acting as positive and negative controls, respectively 26 . 18 F-FDS uptake by A. fumigatus, R. arrhizus, and C. albicans was comparable to its uptake by E. coli, whereas heat-killed pathogens and the Gram-positive bacterium S. aureus did not significantly accumulate the probe ( Fig. 1 and Supplementary Fig. 1). We also tried to visualize A. fumigatus directly under a confocal fluorescence microscopy using nearinfrared fluorescent sorbitol (sorbitol-ZW800-1) and showed that the sorbitol-ZW800-1 was mainly accumulated in germinated A. fumigatus ( Supplementary Fig. 2 Fig. 1 Cellular uptake of radioactivity 1 h after treatment with 18 F-FDS of A. fumigatus, R. arrhizus, C. albicans, S. aureus (−), and E. coli (+) cells and corresponding heat-killed cells. E. coli was the positive control, and S. aureus and heat-killed pathogens were negative controls. Data are expressed as the mean of absolute accumulation activity (Bq) ± SD (per 1 × 10 7 cells) of four or more replicate experiments (A. fumigatus, R. arrhizus, S. aureus, heat killed A. fumigatus, heat killed R. arrhizus, and heat killed C. albicans; n = 5, C. albicans, E. coli, heat killed S. aureus, and heat killed E. coli; n = 4). Statistical significance was calculated using two-tailed Mann-Whitney U tests.
18 F-FDS can detect and differentiate A. fumigatus-infected myositis from sterile inflammation, tumor, and Gram-positive bacterial infection. We next investigated whether 18 F-FDS PET could detect A. fumigatus infection in mouse models. A. fumigatus-infected myositis was generated by inoculation of 1 × 10 8 of A. fumigatus conidia into the right shoulder muscle of immunosuppressed BALB/c mice. The infected muscle was clearly visualized by 18 F-FDS, with its activity retained for 2 h ( Fig. 2a and Supplementary Fig. 3). 18 F-FDS uptake by infected muscle was significantly higher than its uptake by the contralateral normal muscle, with a mean ratio at 2 h of 8.90 ± 1.81 ( Supplementary Fig. 4). Rapid clearance and low non-specific . Mice were inoculated in the right shoulder with 1 × 10 1 to 1 × 10 8 A. fumigatus conidia, and microPET/CT imaging was performed 3 days later. The mice were subsequently sacrificed and the CFU were counted immediately. The correlation was found to be y = 2.5091x + 2.6101 (r = 0.52, P = 0.002). X axis: %ID/g of 18 F-FDS, Y axis: Log (CFU/g) of A. fumigatus. c 18 F-FDS uptake by infected and normal muscles in each subgroup divided by the cutoff number of A. fumigatus conidia (3.1 log CFU/g, ns = not significant, left; n = 20, right; n = 12). Data are expressed as the mean ± SD. Statistical significance was calculated using two-tailed Mann-Whitney U tests.
binding in normal organs resulted in high infection-to-lung, and -liver ratios at 2 h of 9.27 ± 3.32 and 9.33 ± 3.73, respectively (Supplementary Fig. 4). Dynamic 18 F-FDS PET imaging (over 120 min) in mice with A. fumigatus-infected myositis was also performed (Supplementary Movie). 18 F-FDS PET signals could be seen consistently in the infected site even at 120 min, whereas blood pool activities in the heart and liver dissipated after 60 min. This finding indicated that the 18 F-FDS PET signal in A. fumigatus infection is independent of blood pool effects. A. fumigatusinfected myositis was confirmed by pathological examination after PET analysis ( Supplementary Fig. 5).
The limit of detection of 18 F-FDS in A. fumigatus-infected myositis was determined by microPET analysis in mice inoculated with various numbers of A. fumigatus conidia (from 1 × 10 1 to 1 × 10 8 , Fig. 2b). 18 F-FDS uptake was semiquantitatively correlated with the number of Aspergillus conidia in mice with infected muscle (r = 0.52, P = 0.002). Receiver operating characteristic (ROC) curve analysis of the minimum number of A. fumigatus conidia that could be visually detected on 18 F-FDS PET revealed that the optimal cutoff number of A. fumigatus conidia was 3.1 log (CFU/g). The sensitivity, specificity, positive predictive value, negative predictive value, and accuracy of 18 F-FDS PET in the visual detection of A. fumigatus-infected myositis (more than 3.1 log CFU/g of tissue) were 90.0, 83.3, 90.0, 83.3, and 87.5%, respectively. A comparison of 18 F-FDS uptake by infected and normal muscle in each subgroup divided by the cutoff number of A. fumigatus conidia found that, above the cutoff level (≥ 3.1 log CFU/g), 18 F-FDS uptake was significantly higher in infected than in normal muscle (0.89 ± 0.41%ID/g vs. 0.20 ± 0.11%ID/g, P < 0.0001). Below the cutoff level, however, 18 F-FDS uptake did not differ significantly in infected and normal muscle (0.37 ± 0.18%ID/g vs. 0.26 ± 0.07%ID/g, P = 0.0531) (Fig. 2c).
Next, we evaluated whether 18 F-FDS PET could distinguish Aspergillus infection from sterile inflammation, tumor, and Gram-positive bacterial infection in vivo. Live A. fumigatus was injected into the right shoulders and a 10-fold higher number of heat-killed A. fumigatus into the left shoulders of immunosuppressed mice. The right shoulders of other immunosuppressed mice were inoculated with live S. aureus or CT26 colon cancer cells. Infection with S. aureus and sterile inflammation were confirmed by histopathological analysis. Consistent with in vitro uptake data and the in vivo experiments describe above, 18 F-FDS readily concentrated in the Aspergillus-infected shoulders, but not in the shoulders with sterile inflammation, CT26 tumor, and S. aureus infection (Fig. 3a-c), with 18 F-FDS uptake by infected muscle being 3.9-fold higher than by shoulders with sterile inflammation (0.95 ± 0.10%ID/g vs. 0.24 ± 0.01%ID/g, P = 0.0286, Fig. 3d). By contrast, reference imaging showed that 18 F-FDG accumulated not only in Aspergillus-infected shoulders, but also in sterile inflamed, S. aureus-infected, and tumorengrafted shoulders (Fig. 3a-c). 18 F-FDG uptake by shoulders with sterile inflammation was higher than that by shoulders with A. fumigatus infection (8.57 ± 0.77%ID/g vs. 3.63 ± 0.86%ID/g, Fig. 3e).
The in vivo biodistribution study performed 2 h after tracer injection also showed selective concentration of 18 F-FDS in the shoulders with A. fumigatus-infected myositis, with concentrations of 18 F-FDS in these shoulders being 8.4-, 1.8-, 4.0-, 5.5-, and 6.6-fold higher than in blood, lungs, liver, brain, and normal muscle, respectively (Table 1). 18 F-FDS PET can selectively detect and localize A. fumigatus infection in the lung and brain. We then extended our findings to mouse models of IPA. Immunosuppressed BALB/c mice were intranasally inoculated with 1 × 10 8 A. fumigatus conidia or S. aureus. Pulmonary infection with A. fumigatus or S. aureus was confirmed by histopathological analysis (Fig. 4b and Supplementary Fig. 6). 18 F-FDS readily concentrated in the infected lung, but not in the normal lung, and 18 F-FDS PET could clearly visualize Aspergillus infection in the lung parenchyma ( Fig. 4a and Supplementary Fig. 7). More precise examination using highresolution autoradiography clearly demonstrated the specific uptake of 18 F-FDS by A. fumigatus-infected lung tissue (Supplementary Fig. 8). Regions of interest (ROI) analysis showed that uptake of 18 F-FDS by the infected lung was 30.7-fold higher than uptake by normal lung (6.98 ± 2.93%ID/g vs. 0.23 ± 0.03%ID/g, P = 0.0043). 18 F-FDS PET could also distinguish IPA from lung infection by S. aureus, with the latter showing negligible uptake of 18 F-FDS ( Fig. 4a and Supplementary Fig. 9). 18 F-FDS signal intensity was 25.7-fold higher in IPA lungs than in lungs infected with S. aureus (6.98 ± 2.93%ID/g vs. 0.27 ± 0.06%ID/g, P = 0.0095) (Fig. 4c). By contrast, reference imaging with 18 F-FDG could not distinguish IPA from S. aureus lung infection (Supplementary Fig. 10), indicating that 18 F-FDS could identify the etiological agent of pulmonary infection. Although the uptake of 18 F-FDS and 18 F-FDG by A. fumigatus-infected lungs was comparable (6.98 ± 2.93%ID/g vs. 8.38 ± 1.07%ID/g, P = 0.3524, Supplementary Fig. 11), the tracer uptake ratio by infected lung and background tissue (muscle) was significantly higher following 18 F-FDS than 18 F-FDG injection (4.62 ± 1.15 and 0.80 ± 0.07, P = 0.0095, Supplementary Fig. 12). The specificity of 18 F-FDS PET for IPA was further verified by assessing its uptake by lung metastases of malignant melanoma (B16F10 cells) and by lipopolysaccharide (LPS)-induced lung inflammation in mouse models. Pathologically confirmed cancerous lesions and areas of sterile inflammation in the lungs that showed significant 18 F-FDG uptake failed to show significant uptake of 18 F-FDS (Supplementary Figs. [13][14][15]. Because the central nervous system (CNS) is one of the most common sites of Aspergillus infection in immunocompromised hosts, the ability of 18 F-FDS to detect Aspergillus was also evaluated in brain infection. Postmortem histopathologic examination of the mouse cerebrum verified the formation of abscesses and the infiltration of A. fumigatus hyphae in the right cerebrum ( Fig. 5a-c). 18 F-FDS localized to the site of Aspergillus brain infection, but no signal was seen in normal brains ( Fig. 5d and Supplementary Fig. 16). By contrast, 18 F-FDG signal intensities did not differ significantly in infected and normal brains (Fig. 5e), with the uptake ratio between infected brain lesions and surrounding tissue being 12.2-fold higher with 18 F-FDS than with 18 F-FDG (P = 0.0061; Fig. 5f). Overall, these results suggested that 18 F-FDS was a suitable tracer for localization of A. fumigatus brain infection. 18 F-FDS PET can monitor antifungal efficacy in A. fumigatus infection. 18 F-FDS PET could be utilized in patients receiving antifungal treatment to evaluate their microbial burden and to monitor the efficacy of treatment 33,34 . 18 F-FDS PET imaging was therefore performed in mice with A. fumigatus-infected myositis, before and after treatment with voriconazole (40 mg/kg), a drug of choice for the treatment of IA, or fluconazole (20 mg/kg), a negative control. The 18 F-FDS signal, which corresponded to Aspergillus CFU, disappeared after treatment with voriconazole, but not after administration of fluconazole (Fig. 6). The increases in CFU and 18 F-FDS signal in mice receiving fluconazole corresponded to disease progression. Moreover, these mice appeared sick, with body weight loss, ruffled fur, a hunched posture, reduced movement, and earlier death, with two mice dying 8 days and five dying 13 days after treatment, compared with the voriconazole-treated group ( Fig. 6d and Supplementary Fig. 17). Early determination of treatment failure would be helpful in the management of seriously infected patients. infection from other infectious and non-infectious causes. These imaging findings correlated well with the clinical course of IA, thus enabling the effects of treatment to be monitored. 18 F-FDS PET was able to visualize A. fumigatus infections of the lungs, brain, and muscles in mouse models of aspergillosis. The in vivo target-to-background ratios were 31.20 ± 12.82 in IPA (infected/normal lungs), 10.19 ± 2.98 in a model of brain infection (infected/surrounding normal brain), and 8.90 ± 1.81 in a model of myositis (infected/normal muscle). This study also showed that 18 F-FDS uptake was specific for A. fumigatus infection, differentiating this condition from sterile inflammation, tumors, and Gram-positive bacterial (S. aureus) infection in mice. In conventional imaging, such as CT, MRI, and 18 F-FDG PET, pulmonary or CNS aspergillosis is often mistaken for tumors [35][36][37][38] . These diagnoses require additional confirmation by histopathology or culture, but these methods are often too invasive and take several days to get results. In patients with risk factors for IPA, pulmonary infiltration is induced by various causes other than aspergillosis. For example, one study in non-HIV immunocompromised patients found that the etiology of pulmonary infiltrates was infectious in 77% and non-infectious in 23% of patients, with the most frequent pathogens being A. fumigatus (23%), S. aureus (14%), and Pseudomonas aeruginosa (10%) 39 . Therefore, distinguishing A. fumigatus infection from non-infectious or other infectious causes is critical. The present study verified that 18 F-FDS uptake could differentiate A. fumigatus infection from other infectious and non-infectious causes. By contrast, 18 F-FDG uptake could not differentiate Aspergillus infection from sterile inflammation, CT26 tumor, and S. aureus infection. Especially in the brain aspergillosis model, the infected area was highly localized and visualized by 18 F-FDS, not by 18 F-FDG. One of the major advantages of 18 F-FDS as a tracer is that it can be synthesized rapidly from 18 F-FDG through simple reduction, which can be easily performed either on-site or at a nearby cyclotron. Therefore, 18 F-FDS PET imaging has potential application in the early diagnosis of IA in clinical settings. Even if 18 F-FDS cannot be prepared as quickly as needed, its high accuracy suggests that 18 F-FDS would be a promising tracer for more delayed diagnosis. Moreover, empirical antifungal therapy, which is usually administered to patients with suspected IA, could be adjusted to definitive therapy based on the results of 18 F-FDS PET scanning. 18 F-FDS PET scanning after voriconazole treatment of mice with A. fumigatus-infected myositis showed that uptake diminished over time. These imaging findings correlated well with the clinical courses of these mice, in that they survived until the end of treatment (16 days post-infection), with improved appearance and increased physical activity. Evaluation of the efficacy of antifungal agents may be difficult in immunocompromised patients with IA. Because the clinical signs and symptoms of IA can be non-specific, use of these clinical parameters to characterize treatment response may lead to erroneous interpretations. Clinical manifestations suggesting ongoing inflammatory reactions can indicate treatment failure in patients receiving antifungal therapy, or they may be related to other factors, including concomitant infection, underlying medical conditions, and immune reconstitution inflammatory syndrome 40 . Conventional radiology, such as CT and MRI, cannot monitor therapeutic responses because they reveal structural abnormalities and inflammation, which do not subside rapidly. Thus, the ability of 18 F-FDS to monitor the course of treatment suggests that this method will play a certain role in clinical decision making in patients with IA. c Quantification of the in vivo 18 F-FDS and 18 F-FDG PET images (ns = not significant). Data are expressed as the mean ± SD. Statistical significance was calculated using two-tailed Mann-Whitney U tests.
18 F-FDS PET imaging in mice with IPA revealed high background uptake, which had not been previously observed in other studies using animals infected with E. coli or Klebsiella pneumoniae 26,27 . In the most recent study on 18 F-FDS uptake in Aspergillus spp., dynamic PET/CT imaging showed that the radioactivity of 18 F-FDS in the blood appeared to be maintained longer in A. fumigatus infected mice than in healthy control mice, which might be due to hemodynamic change, such as systemic vasodilation caused by serious infection 41 . A blood pool effect of 18 F-FDS PET was observed in human patients, suggesting that our findings may also be associated with a blood pool effect 34 . However, if the increased background uptake of 18 F-FDS was solely related to enhanced blood pool effect, which is the manifestation of sepsis, similar uptake should have been observed in mice with S. aureus lung infection that were as sick as IPA mice. Even if pathogen-specific uptake and an increased blood pool effect coexist in IPA mice, the potential of 18 F-FDS PET imaging should not be overlooked, given that in diverse IA models the target-to-background ratio is high in brain and muscle, as well as in lung (4.62 ± 1.15), which is in contrast to 18 F-FDG PET imaging (0.80 ± 0.07, Supplementary Fig. 12). An in vitro uptake assay demonstrated that 18 F-FDS uptake in A. fumigatus was as high as in E. coli when it was normalized to 1 × 10 7 conidia or CFU. However, when in vitro 18 F-FDS uptake of microbes was normalized to protein, significant reduction of uptake was observed in fungi in comparison with E. coli (Supplementary Fig. 1). As fungi and bacteria have very different growth patterns, dynamics, and metabolism, determination of a universal indicator to normalize the biomass for both is difficult 41 . In broth media, not all conidia of A. fumigatus are activated and germinated to form hyphae, and some remain as dormant conidia, even after relatively long incubation periods 42 . As hyphae take up more sorbitol than conidia 41 , if dormant conidia remain in the culture media, the uptake value might be underestimated when normalized with protein. In a further experiment with near infrared fluorescence microscopy, we were able to directly observe accumulation of fluorescent sorbitol (sorbital-ZW800-1) in germinated A. fumigatus in comparison with conidia ( Supplementary Fig. 2). In most in vivo infected tissue, A. fumigatus is observed as well-developed hyphae, as germination is an essential phase in invasive Aspergillus infection 4 , and we also observed well-developed hyphae in our histological findings (Figs. 4 and 5 of the manuscript and Supplementary Fig. 5). We subsequently observed significantly higher 18 F-FDS uptake in the infected tissue than in the background tissue on PET imaging. Therefore, we speculate that because of differences in the growth of A. fumigatus between biotic and abiotic environments, in vitro uptake might be underestimated when normalized according to the amount of protein.
This study had several limitations. Because 18 F-FDS also accumulates in Enterobacterales 26 , 18 F-FDS PET imaging could not differentiate A. fumigatus infection from Enterobacterales infection. Clinically, however, IA is mainly diagnosed in immunocompromised hosts and generally manifests as lung, sinus, and CNS infections, whereas Enterobacterales mainly cause intraabdominal infection regardless of host immune status. The combination of 18 F-FDS PET and appropriate clinical information may have a role in the differential diagnosis and localization of IA. Testing also showed that 18 F-FDS was taken up in vitro by cells of other fungi, including R. arrhizus and C. albicans. 18 F-FDS PET imaging of these fungi has not been evaluated in vivo to date. In contrast to mold infection, candida is an infrequent cause of pulmonary, brain, and sinus infections, and is relatively easy to diagnose microbiologically. Infection with Rhizopus, one of the most frequent causes of invasive mucormycosis, is usually fatal and may be difficult to distinguish from IA. Mucormycosis is increasingly reported related to COVID-19. Although most of these patients had rhino-orbital cerebral disease, some also had pulmonary disease 43 . Additional studies are needed to evaluate the usefulness of 18 F-FDS PET imaging in the diagnosis of Rhizopus infection.
Our study shows that 18 F-FDS PET can monitor A. fumigatus infection, a leading cause of death in immunocompromised patients. 18 F-FDS has favorable aspects for imaging IA, such as high selectivity for infected tissue and rapid clearance from other organs, leading to superior image quality. In addition, 18 F-FDS is expected to be cost effective, owing to the inexpensive and easy radiosynthesis process. Furthermore, the radioactive decay product of 18 F-FDS, sorbitol, has been approved as a food worldwide, including by the U.S. Food and Drug Administration and the European Food Safety Authority. Thus, 18 F-FDS may be an ideal Data are expressed as the mean ± SD. Statistical significance was calculated using two-tailed Mann-Whitney U tests.
candidate for molecular imaging of IA in clinical settings, providing a rapid, noninvasive, and accurate diagnostic tool to identify and localize Aspergillus infections, guide the selection of mold-active antifungals, and improve patient outcomes.
In summary, the present study showed that 18 F-FDS has a potential as a tool for non-invasive detection of A. fumigatus infections in vivo. The selective concentration and prolonged retention of 18 F-FDS in regions of A. fumigatus infection with low background activity suggest that this radiotracer could be used as a PET molecular imaging agent to obtain high quality images in the diagnosis of A. fumigatus infection. The excellent and unique characteristics of 18 F-FDS warrant its further investigation for clinical translation.

Methods
Synthesis of 19 F-FDS. 19 F-FDS was prepared by a modification of a previously described method, involving the reduction of 19 F-FDG by sodium borohydride in water 32 . Sodium borohydride (10 mg, 0.265 mmol) was added to a solution of 19 F-FDG (10 mg, 0.055 mmol) in water and the reaction mixture was stirred at 45°C for 15 min. The mixture was adjusted to pH 6.5-7.5 and passed through a Sep-Pak Alumina N cartridge. The filtrate was evaporated and 19 F-FDS was obtained at a yield of 8.9 mg (87.9%). Parameters In vitro uptake assays. Escherichia coli (ATCC 25922), S. aureus (ATCC 29213), A. fumigatus (isolated from a sputum sample from a patient admitted to the oncology ward of Chonnam National University Hwasun Hospital), R. arrhizus (clinical isolate), and C. albicans (SC5314) were used for in vitro uptake assays. E. coli and S. aureus were aerobically grown in Lysogeny Broth (LB) overnight at 37°C; 1 mL of each sample was transferred to fresh LB and incubated for another 2 h at 37°C. Colonies were counted and 1 × 10 7 CFU of E. coli and S. aureus were prepared for probe uptake assay. C. albicans was cultured on yeast extract peptone dextrose (YPD) plates at 37°C for 48 h, harvested using sterile phosphate-buffered saline (PBS), and suspended to a final concentration of 1 × 10 7 CFU. A. fumigatus and R. arrhizus were cultured on potato dextrose agar plates for 5-7 days at 30°C, and colonies were harvested using saline containing 0.1% (v/v) Tween 20. The suspensions were vortexed to release the conidia and filtered through a 40 μm cell strainer to remove clumps and hyphae. Colonies were counted using a hemocytometer, and 1 × 10 7 A. fumigatus conidia were prepared for probe uptake assay. The control consisted of A. fumigatus heat-killed in an autoclave at 121°C for 15 min 44 . Probe uptake assays were performed by incubating bacterial or fungal cultures with 18 F-FDS (0.74 MBq/mL) at 37°C with rapid agitation (200 rpm). Bacteria or fungi were filtered using micro centrifuge tube filters and washed three times with PBS. The radioactivity of each supernatant and pellet was measured using a gamma counter.
To present the Bq/μg of protein in in vitro uptake studies, 1 × 10 6 CFU of cell pellets suspended in 0.5 mL of Y-PER TM yeast protein extraction reagent (Thermo-Fisher Scientific, 78991) or B-PER TM bacterial protein extraction reagent (Thermo-Fisher Scientific, 78243) were mixed with Protease Inhibitor Cocktail Solution (GenDEPOT, P3100-010) and transferred to 2 mL reinforced tubes (Precellys, P000943-LYSK0-A.O) containing 200 μL dry volume of 212-300 μm glass beads (Sigma-Aldrich, G9143). The mixture was incubated on ice for 30 min, and the cells were disrupted by five 65 s blasts of vigorous vortexing, interspersed with 2 min incubations of the tube on ice. Following the final vortexing, the beads were allowed to settle and each supernatant was decanted into a fresh tube. Protein concentrations were quantified using BCA™ Protein Assay kits (Thermo-Fisher Scientific, 23225).
Confocal microscopy. A. fumigatus was cultured on potato dextrose agar plates for 5-7 days at 30°C, and colonies were harvested using PBS containing 0.1% (v/v) Tween 20. The suspensions were vortexed to release the conidia and filtered through a 40 μm cell strainer to remove clumps and hyphae. Conidia were counted using a hemocytometer, and 1 × 10 6 A. fumigatus conidia and 50 µM of sorbitol-ZW800-1 were co-cultured for up to 48 h in 1 mL of YPD broth in 24-well plates at 37°C without agitation. Sorbitol-ZW800-1 was prepared as described previously 45 . Briefly, to prepare sorbitol-ZW800-1, ZW800-1 NHS ester (1 mmol, 1 mg) was conjugated with amino-sorbitol (1.5 mmol, 0.27 mg) using N,N-diisopropylethyl amine (5 mmol, 0.65 mg) in DMSO (2 mL) at room temperature for 1 h. The crude mixture was purified by a preparative HPLC system. Co-cultured A. fumigatus were observed at 20 and 48 h using laser scanning confocal microscopy (LSM 800, ZEISS, Germany). Before microscopic examination, cultured A. fumigatus were washed twice, re-suspended in PBS, and transferred to a confocal dish. Fluorescence of ZW800-1 (red) was obtained using laser filter excitation at 561 nm and emission at 650-750 nm. Image acquisition and analysis were performed using ZEN 2.6, blue edition (ZEISS, Germany).
Animal models and biodistribution studies. Animal care, animal experiments, and euthanasia were performed in accordance with protocols approved by the Chonnam National University Animal Research Committee and the Guide for the Care and Use of Laboratory Animals.
BALB/c mice (female, 8 weeks old) were used for muscle and lung infection; C57BL/6 mice (female, 6 weeks old) were used for brain infection and B16F10 (mouse melanoma cell line) lung metastases; and C57BL/6 (male, 8 weeks) mice were used for LPS-induced acute lung injury. A. fumigatus-infected myositis models have been described 46 . Briefly, BALB/c mice were immunosuppressed by intraperitoneal injections of cyclophosphamide (200 mg/kg) 4 and 1 days prior to inoculation and a subcutaneous injection of cortisone acetate (125 mg/kg) 1 day prior to inoculation, followed by inoculation of the right shoulder muscle with 1 × 10 8 A. fumigatus conidia. 18 F-FDS scans were performed 72 h after infection.
To evaluate the correlation between 18 F-FDS uptake and number of cells, mice were inoculated in their right shoulders with 1 × 10 1 to 1 × 10 8 A. fumigatus conidia. Following 18 F-FDS imaging, the mice were sacrificed and their infected right shoulder muscles were weighed and homogenized in 5 mL of sterile saline. Homogenates were serially diluted and spread on potato dextrose agar plates, which were incubated at 30°C for 48 h. A. fumigatus CFU were counted, with the data expressed as log CFU/g of tissue. The minimum number of A. fumigatus conidia that could be visually identified on 18 F-FDS PET was determined by ROC curve analysis. 18 F-FDS uptake by infected and normal muscle was compared in each subgroup, divided by the cutoff number of A. fumigatus conidia.
The S. aureus-infected myositis model was generated by injecting 1 × 10 8 CFU S. aureus into the right shoulder muscles of sedated 8-week-old female BALB/c mice 29 . Sterile inflammatory myositis was induced by injection into the right shoulder muscles of heat-killed conidia 44 . To generate tumor xenografts, CT26 (mouse colon carcinoma) or B16F10 cells were cultured in DMEM supplemented with 10% fetal bovine serum (FBS) at 37°C in a 5% CO 2 incubator. CT26 cells (1 × 10 6 ) were injected into the right shoulder muscles of female, 6-week-old BALB/c mice. The lung metastases models were generated by i.v. injection of B16F10 cells (2 × 10 5 ) 47 . After 10 days, the mice were subjected to microPET imaging.
To generate A. fumigatus-infected lungs, BALB/c mice were immunosuppressed by intraperitoneal injection of cyclophosphamide (150 mg/kg) 4 days prior to inoculation and subcutaneous injection of cortisone acetate (250 mg/kg) 1 day prior to inoculation, followed by intranasal inoculation with 1 × 10 8 A. fumigatus conidia 48 . S. aureus-associated pneumonia was induced in sedated 8-week-old female BALB/c mice by intranasal instillation of 40 µl of PBS containing 2 × 10 8 CFU S. aureus 49 . 18 F-FDS scans were performed 48 h after infection unless otherwise indicated. A mouse model of acute sterile lung inflammation 50 was induced by intraperitoneal injection of LPS (20 mg/kg, E. coli 055:B5, Sigma-Aldrich; St. Louis, MO) into C57BL/6 mice, with 18 F-FDS imaging performed 4 h after LPS injection. For brain infection, C57BL/6 mice were immunosuppressed with 200 mg/kg of cyclophosphamide 4 days prior to infection. On the day of brain infection, the mice were anesthetized with a 4:1 mixture of alfaxalone:xylazine and fixed in a stereotactic frame. The midline scalp was dissected and a small hole (3 mm caudal and 3 mm lateral to the right of the bregma) was drilled into the skull of each mouse. Using a 10 μL Hamilton syringe, a suspension of 1 × 10 5 A. fumigatus conidia in 3 µl saline was injected through the hole 3.5 mm deep into the brain of each mouse. Immediately after retraction of the needle, the hole in the skull was sealed with bone wax and the scalp was sutured 51 . 18 F-FDS scans were performed 24 h, 48 h, and 72 h after infection.
The biodistribution of 18 F-FDS in different organs was assessed in mice with A. fumigatus-infected myositis 2 h after intravenous injection of 7.4 MBq of radiotracer. After sacrifice, the blood, heart, lungs, liver, spleen, stomach, intestines, kidneys, pancreas, normal muscles, bones, brain, skin, and infected muscles were extracted from the mice and weighed. The radioactivity of each organ was determined using a gamma counter, with these measurements of radioactivity normalized relative to the weight of each organ sample and the amount of radioactivity injected to obtain the % ID/g (ID: injected dose).
Antifungal efficacy. To measure the efficacy of antifungal treatment, mice infected with A. fumigatus on the right shoulder were intraperitoneally injected with 40 mg/kg of voriconazole (Pfizer) every 24 h from day 3 to day 16 post-infection. Control mice were similarly injected with 20 mg/kg fluconazole (Pfizer) every 24 h from day 3 to day 16 post-infection. All mice also received intraperitoneal injections of cyclophosphamide (200 mg/kg) on days 6, 9, 12, and 15 postinfection. MicroPET scans were performed on post-infection days 3 (before treatment), 11 (after 8 days of treatment), and 16 (after 13 days of treatment). Following intravenous injection of 7.4 MBq 18 F-FDS, microPET images were acquired for 10 min at 2 h.
Small animal PET/CT examination. PET images were obtained 48 h after inoculation of A. fumigatus or S. aureus into mouse lungs; 24, 48, and 72 h after inoculation of A. fumigatus into brain; and 72 h after inoculation of A. fumigatus or S. aureus into muscle. All images were acquired using a high-resolution small animal PET-SPECT-CT scanner (Inveon, Siemens Medical Solutions). One or two hours after intravenous injection of 7.4 MBq 18 F-FDS or 18 F-FDG, micro-PET/CT images were acquired for 10 min. To compare 18 F-FDS and 18 F-FDG images of CT26 tumors, mice with these tumors were injected with 18 F-FDS 1 day after 18 F-FDG microPET imaging. ROI were drawn manually over the infected organs, normal organs, tumor, and areas of inflammation on decay-corrected coronal images, using CT images as a guide. Acquired images were reconstructed with a three-dimensional ordered subset expectation maximization (OSEM3D) algorithm. MicroPET image analysis was performed with PMOD software (PMOD Technologies Ltd, Zürich, Switzerland). Radiotracer uptake was expressed as %ID/g.
High-resolution autoradiography. Mice with A. fumigatus-infected lungs were injected intravenously with 111 MBq of 18 F-FDS and sacrificed. Their lungs were extracted, frozen in isopentane, and cut into 20 μm sections using a cryostat. The sections were thaw-mounted onto glass microscope slides, which were exposed to the phosphor screen overnight and scanned in a Typhoon FLA 9500 scanner (GE Healthcare, USA). Photographs of the sections were recorded with a digital camera.