Imaging Diverse Pathogenic Bacteria In Vivo with 18F-Fluoromannitol PET

Visual Abstract

Infectious disease remains the main cause of morbidity and mortality throughout the world. Of growing concern is the rising incidence of multidrug-resistant bacteria, derived from various selection pressures. Many of these bacterial infections are hospital-acquired and have prompted the Centers for Disease Control and Prevention in 2019 to reclassify several pathogens as urgent threats, its most perilous assignment. Consequently, there is an urgent need to improve the clinical management of bacterial infection via new methods to specifically identify bacteria and monitor antibiotic efficacy in vivo. In this work, we developed a novel radiopharmaceutical, 2-18 F-fluoro-2deoxy-mannitol ( 18 F-fluoromannitol), which we found to specifically accumulate in both gram-positive and gram-negative bacteria but not in mammalian cells in vitro or in vivo. Methods: Clinical isolates of bacteria were serially obtained from wounds of combat service members for all in vitro and in vivo studies. Bacterial infection was quantified in vivo using PET/CT, and infected tissue was excised to confirm radioactivity counts ex vivo. We used these same tissues to confirm the presence of bacteria by extracting and correlating radioactive counts with colony-forming units of bacteria. Results: 18 F-fluoromannitol was able to differentiate sterile inflammation from Staphylococcus aureus and Escherichia coli infections in vivo in a murine myositis model using PET imaging. Our study was extended to a laceration wound model infected with Acinetobacter baumannii, an important pathogen in the nosocomial and battlefield setting. 18 F-fluoromannitol PET rapidly and specifically detected infections caused by A. baumannii and several other important pathogens (Enterococcus faecium, S. aureus, Klebsiella pneumoniae, A. baumannii, Pseudomonas aeruginosa, and Enterobacter spp.). Importantly, 18 Ffluoromannitol PET was able to monitor the therapeutic efficacy of vancomycin against S. aureus in vivo. Conclusion: The ease of production of 18 F-fluoromannitol is anticipated to facilitate wide radiopharmaceutical dissemination. Furthermore, the broad sensitivity of Inf ection is responsible for the highest morbidity and the third most deaths among all human diseases worldwide (1). Most health-care-associated infections in the United States arise from several common pathogens, including Staphylococcus aureus, Acinetobacter baumannii, Pseudomonas aeruginosa, and those of the Enterobacteriaceae family (Escherichia coli and Salmonella spp., among others). The rising trend of antimicrobial resistance, compounded by a growing population of immunocompromised individuals (HIV/AIDS, chemotherapy, organ transplantation, diabetes) creates an enormous economic strain on the U.S. health-care system, with estimates ranging from $28 billion to $45 billion annually (2). Current estimates project that drug-resistant infections will become the leading cause of global death, surpassing cancer-associated mortality by 2050 (3). The Centers for Disease Control and Prevention has recently listed carbapenem-resistant Acinetobacter and Enterobacteriaceae, extended-spectrum b-lactamase-producing Enterobacteriaceae, multidrug-resistant P. aeruginosa, methicillin-resistant S. aureus, and others as urgent or serious threats to human health (4). Carbapenem-resistant A. baumannii alone was responsible for 8,500 hospitalizations, 700 deaths, and $281 million in U.S. healthcare costs in 2017 (4). Carbapenem-resistant A. baumannii infections are particularly problematic for patients who have comorbidities or are immunocompromised (5); however, A. baumannii-associated infections are also well-described complications of severe combatrelated injuries in military service members (6). Accordingly, there is an urgent need to improve the diagnosis and treatment of bacterial infection.
Traditional approaches to diagnosing infection include obtaining a biopsy sample from tissue or blood and subsequently culturing pathogens in media to identify an organism. Bacterial cultures from tissue biopsy specimens remain the gold standard for confirming the presence, identity, and drug sensitivity of a microorganism; however, deep-seated infections that are difficult to access or identify often rely on noninvasive imaging techniques based on changes in anatomy or tissue morphology. The most common anatomic imaging modalities used, such as CT and MRI, are frequently nonspecific for delineating active infection from sterile inflammatory disease. Nuclear medicine uses labeled leukocytes ( 99m Tc-or 111 In-oxine) (7) and 67 Ga-citrate scintigraphy (8), which rely on indirect measurements of leukocyte recruitment to an area of interest. PET imaging with 18 F-FDG is increasingly used; however, none of these imaging techniques can distinguish active infection from cancer or inflammation. Consequently, current clinically available imaging techniques are not adequately specific to diagnose deep-seated infection.
To address this challenge, many recently developed radiopharmaceuticals seek to exploit various bacteria-specific signatures such as metabolism (9)(10)(11), cofactor biosynthesis (12,13), and labeled antibiotics (14,15). Despite these scientific advances, a dire need persists for imaging agents that meet the challenges of clinical infectious disease practice; the ideal agent should possess broad bacterial strain sensitivity, have optimal pharmacokinetics (rapid target engagement, clearance of nonspecific signals to promote contrast), and be widely deployable and available for clinical use.
The phosphoenolpyruvate-dependent sugar phosphotransferase system catalyzes phosphorylation of incoming sugar substrates, with concomitant translocation across the cell membrane, and is widely found in bacteria (16)(17)(18). Because of this metabolic signature, 3 H-and 14 C-D-mannitol analogs were recently evaluated in a panel of pathogens (19). We hypothesized that a positron-emitting analog of mannitol, 2-18 F-fluoro-2-deoxy-mannitol ( 18 F-fluoromannitol), would be a specific precursor for bacterial metabolism and, subsequently, a suitable imaging agent for in vivo use with PET. Here, we report a simple, widely deployable radiosynthesis of 18 F-fluoromannitol and demonstrate that this imaging agent possesses broad-spectrum bacterial sensitivity both in vitro and in vivo using clinical isolates of bacteria from combat wounds in military service members. Moreover, we demonstrate that 18 F-fluoromannitol can quantify antimicrobial efficacy in vivo.

MATERIALS AND METHODS
Manual Radiosynthesis of 18 F-Fluoromannitol 18 F-fluoromannitol was synthesized from commercially available cyclotron-derived 18 F-fluoride ions and isolated in a radiochemical yield of 23% 6 2% (end of synthesis) with an estimated molar activity of 5.5 6 0.37 GBq/mmol (n 5 14). Detailed radiosynthetic procedures are described in the supplemental methods (supplemental materials are available at http://jnm.snmjournals.org).

Wound Infection Model
C57BL/6 mice (male, 5-6 wk old) were used for all experiments. A 3-mm laceration in the dorsal fascia was injected with 50 mL (typical inoculations were 10 6 CFUs) of A. baumannii into the open wound, and the infection was allowed to develop for 6 h (11) before imaging.

PET/CT Imaging
For all studies, 5.5 6 1.8 MBq were injected via a lateral tail-vein catheter. After injection, mice were imaged by dynamic (60 min) or static (45-60 min) PET acquisition. All scans were immediately followed by a 10-min CT scan for attenuation correction and anatomic coregistration. Afterward, the mice were euthanized for biodistribution studies and CFU analysis when applicable. g-counting of harvested tissue was performed using an automatic g-counter (Hidex). Detailed protocols are described in the supplemental methods.

Computation, Registration, and Quantification of Parametric PET Maps
Parametric PET maps of the total rodent body were generated and computed as previously described (21). Net influx rate maps were computed and coregistered with CT images using PMOD (version 3.9, PMOD Technologies). The regional average net influx rate was quantified and correlated with CFUs. Detailed protocols are described in the supplemental methods.

Statistical Methods
Quantitative data are expressed as mean 6 SEM unless otherwise noted. Means were compared using 1-way ANOVA or, for multiple comparisons, 2-way ANOVA. The Mann-Whitney U test was used to test significant differences in SUV comparisons over time (dynamic imaging). P values smaller than 0.05 were considered statistically significant.
Radiochemical synthesis of 18 F-fluoromannitol is a straightforward 3-step, 2-pot process using commercially available precursor. Radiosynthesis produces 18 F-fluoromannitol in high radiochemical yield and purity, which are easily determined by radio-high-performance liquid chromatography. Molar activity was measure of 3 radiosyntheses. Intermediates were verified using fully characterized 19 F isotopic standard and matched to high-performance liquid chromatography retention times. FDM 5 2-18 F-fluoro-2-deoxy-mannose; FMtl 5 fluoromannitol.

Characterization of 18 F-Fluoromannitol In Vitro
Over time, both S. aureus and E. coli readily incorporated 18 Ffluoromannitol ( Fig. 2A) but not heat-killed bacteria, demonstrating metabolic specificity of bacteria for 18 F-fluoromannitol. We next evaluated accumulation of 18 F-fluoromannitol in a broad panel of bacterial strains (Supplemental Table 1). All strains tested, except for P. aeruginosa and Enterococcus faecium, showed rapid and significant accumulation (Fig. 2B). Coincubation of 18 F-fluoromannitol with D-mannitol in S. aureus cultures demonstrated target specificity and that accumulation of 18 F-fluoromannitol is not concentration-dependent (Fig. 2C); concentrations of at least 10 mg/mL of D-mannitol blocked 18 F-fluoromannitol accumulation in bacteria. We also compared the accumulation of 18 F-fluoromannitol in S. aureus (gram-positive) and E. coli (gram-negative) cultures against 18 F-FDG, the current workhorse of nuclear medicine, and 2-deoxy-2-18 F-fluorosorbitol ( 18 F-FDS), which has demonstrated high specificity for Enterobacteriaceae organisms (10). As anticipated, 18 F-fluoromannitol accumulated in both E. coli and S. aureus, and in S. aureus this accumulation was significantly higher than that of 18

F-Fluoromannitol Characterization In Vivo
We used a murine myositis model of musculoskeletal infection (10)(11)(12)14,20,23) to determine whether 18 F-fluoromannitol can differentiate sterile inflammation from infection in vivo by inoculating the right triceps brachii with a live strain of bacteria and the left triceps brachii with a 10-times quantity of heat-killed bacteria to generate an inflammatory response. 18 F-fluoromannitol accumulated specifically in the site of infection in both gram-positive and gram-negative strains (Fig. 3A). 18 F-FDG was predictably unable to distinguish infection from inflammation, consistent with prior reports (10)(11)(12)14,20,23), but did serve as a valuable positive control. Dynamic imaging revealed rapid accumulation and significant differences in PET signal in as little as 5 min after 18 F-fluoromannitol injection.
To quantify PET signal, we generated volumes of interest in the upper limbs of mice using CT for anatomic localization. 18 F-fluoromannitol displayed a 3.5-fold increased SUV (summed frames 45-60 min after injection) compared with the contralateral site of inflammation (Supplemental Fig. 8). 18 F-FDG could not show significant differences in SUV between sites of infection and inflammation. After the scans, we excised both triceps brachii to confirm the PET data using g-counting, which confirmed the increased PET signal in the infected tissue compared with inflamed tissue (Supplemental Fig. 9). Biodistribution studies were performed in successive cohorts of mice over 3 h to determine the dosimetry of 18 F-fluoromannitol ( Fig. 3B; Supplemental Fig. 10). The kidneys and bladder demonstrated the highest nonspecific accumulation of 18 F-fluoromannitol, consistent with PET imaging data. We also correlated static PET SUV with bacterial CFUs from excised tissue to determine the sensitivity of 18 F-fluoromannitol ( Fig. 3C;  Supplemental Fig. 11) and found that 18 F-fluoromannitol can reliably detect as little as 5 log 10 (CFUs/mL) of bacteria in vivo by SUV. We also investigated whether parametric mapping (21) can increase the bacterial sensitivity of 18 F-fluoromannitol in vivo (as net influx rate is a quantitative measure of the rate of uptake in tissue (24)), rather than SUV (which is semiquantitative and cannot delineate signal from blood pool contamination and tissue). Parametric imaging improved the bacterial sensitivity of 18 F-fluoromannitol by roughly 20-fold (log 10 CFUs 5 1.3; 1.7 3 10 6 improved to 7.0 3 10 5 CFUs) (Figs. 3C and 3D; Supplemental  Fig. 12).
We also investigated the sensitivity of 18 F-fluoromannitol compared with 18 F-FDS in a mixed infection (polymicrobial) model. 18 F-FDS has shown remarkable specificity for Enterobacteriaceae in vivo but has shown limited to no sensitivity toward gram-positive and other gram-negative organisms. Mice were inoculated with live E. coli (8.4 3 10 6 CFUs) and S. aureus (8.8 3 10 6 CFUs) in the right and left triceps brachii, respectively. No significant differences in SUV (P 5 0.19) were observed between E. coli and S. aureus infection in the same animal (Figs. 4A and 4B) with 18 Ffluoromannitol. Importantly, 18 F-fluoromannitol accumulation was significantly higher than 18 F-FDS accumulation in S. aureus infection (P , 0.001). 18 F-FDS demonstrated high specificity for E. coli, compared with S. aureus (P 5 0.007); however, no significant differences in E. coli SUV were evident between 18 F-fluoromannitol and 18 F-FDS (P 5 0.11). Postmortem g-counting of tissues confirmed 18 F-fluoromannitol uptake in both S. aureus and E. coli, whereas 18 F-FDS accumulated only in E. coli-infected muscle (Supplemental Fig. 13). In sum, our imaging data show that 18 F-fluoromannitol accumulates in both gram-positive and gramnegative organisms and is of adequate sensitivity to serve as an in vivo broad-spectrum imaging tool for infection.

Imaging Wound Infection with 18 F-Fluoromannitol PET
The emergence of carbapenem-resistant A. baumannii has rendered clinical management of A. baumannii infections difficult to impossible (25) in some cases. The urgency to improve management of A. baumannii infections prompted us to investigate whether 18 F-fluoromannitol can detect A. baumannii in a laceration wound model. Mice were inoculated with A. baumannii (9.4 3 10 6 CFUs) through a small incision in the dorsal fascia and imaged using PET/CT. 18 F-fluoromannitol accumulated specifically in the infected wounds of mice ( Fig. 5; Supplemental Fig. 14), demonstrating nearly a 6-fold increase in SUV (7.0 log 10 CFUs) compared with a 10-times quantity of heatkilled bacteria in a nearby wound.

Quantifying Antimicrobial Efficacy In Vivo
The growing incidence of antimicrobial resistance in many bacterial pathogens is a serious concern because treatment failure is a major threat to global health (26). Inappropriate antibiotic use is also the primary driver of antibiotic resistance (27), which also places undue risk on patients for adverse events such as allergic reactions and Clostridium difficile infection. Thus, it is imperative to optimize the management of infection and use of antibiotics. We investigated whether 18 F-fluoromannitol can quantify the efficacy of antibiotic therapy in vivo. Mice were inoculated with S. aureus in the right triceps brachii and imaged with 18 F-fluoromannitol 8 h after infection, before initiation of vancomycin treatment (100 mg/kg every 8 h, intraperitoneally), and subsequently were imaged at 24 and 72 h after treatment. The PET signal diminished over the course of treatment, correlating closely with CFU burden (Fig. 6A). We next investigated the accumulation of 18 F-fluoromannitol in a panel of bacterial isolates from infected combat wounds of military service members (Fig. 6B). 18 F-fluoromannitol demonstrated broad accumulation in S. aureus and A. baumannii but did not show appreciable accumulation in P. aeruginosa. Taken together, these findings indicate that 18 Ffluoromannitol can be used as an effective tool to image a variety of clinically relevant pathogens.

DISCUSSION
Mortality-associated infection disproportionately affects populations with strained access to health care (28); thus, a critical metric for any imaging agent is that it be easily disseminated. The use of 18 F ensures that the isotope is regularly available from cyclotron production, and the half-life (109.5 min) facilitates widespread distribution. The radiosynthesis of 18 F-fluoromannitol (Fig. 1) is a straightforward 3-step reaction; the first 2 steps were intentionally designed to model the radiosynthesis of 18 F-FDG, followed by seamless sodium borohydride reduction (29). All purifications are cartridge-based and facilitate automation on any radiosynthesizer, promoting robust access to 18 F-fluoromannitol.
Clinical management of infection typically commences with empiric antibiotic therapy using broad-spectrum agents, often combined with a targeted antimicrobial. Treatment generally continues until biopsy or culture reveals the causative organism; however, treatment may continue in lieu of positive identification. The inability to rapidly delineate bacterial infection promotes unnecessary exposure to antibiotics, contributing to the rising incidence of antimicrobial resistance mechanisms and morbidities associated with antibiotic therapy (30,31). 18 Ffluoromannitol shows rapid and specific accumulation in bacteria in vivo in several clinically relevant pathogens. Despite the seeming disparity in accumulation of 18 F-fluoromannitol in E. coli compared with 18 F-FDS in vitro (Fig. 2D), no significant differences (P 5 0.11) in tracer accumulation were observed in vivo in E. coli (Fig. 4). This important observation highlights that although in vitro assays play fundamentally important roles in preliminary characterization and validation, it is imperative that other characteristics, such as pharmacokinetics, not be overlooked when evaluating the candidacy of a novel radiopharmaceutical for imaging. Collectively, the minimal nonspecific accumulation and radioactive dose of 18 F-fluoromannitol in mammalian tissue suggest that this agent is well poised for clinical studies on anatomic localization of a variety of infections.
Imaging can realistically play a complementary role in managing several clinical applications of infection with diverse etiologies. However, the complementary role imaging will play is ultimately limited to the in vivo sensitivity of the agent (CFUs/mL). Several radiopharmaceuticals have been studied, including glucose ( 18 F-FDG) (32,33), sorbitol ( 18 F-FDS) (10,34), and maltose ( 18 F-fluoromaltose, 18 F-fluoromaltotriose) (11). 18 F-FDS has shown adequate sensitivity to infection in both preclinical models and human disease; however, this agent is limited to the detection of Enterobacterales. Maltose-derived radiopharmaceuticals demonstrated improved strain coverage that includes P. aeruginosa and S. aureus; however, the sensitivity of these agents for clinically relevant concentrations of bacteria beyond E. coli remains uncertain. Other imaging agents, such as those targeting folate biosynthesis (12)(13)(14) or transpeptidases (20,23), report limited (10 8 CFUs) or unknown sensitivity. 18 F-fluoromannitol was able to reliably detect 10 5 CFUs in vivo using the clinical standard SUV, which is of sufficient sensitivity for detecting an abscess (3). Furthermore, the sensitivity of 18 Ffluoromannitol did not diminish between E. coli and S. aureus, suggesting that sensitivity is not dependent on a specific genus or family of bacteria.
The Centers for Disease Control and Prevention estimates that approximately 30% of prescribed antibiotics are unnecessary (27), and it is alarming that inappropriate use of antibiotics is the primary driver for the development of antibiotic resistance mechanisms. Bacterial CFUs were shown to correlate with PET SUV during vancomycin treatment using 18 F-fluoromannitol imaging (Fig. 6A). In addition, 18 F-fluoromannitol demonstrates indistinguishable accumulation in E. coli and S. aureus in vivo. 18 F-fluoromannitol is well positioned to serve as a valuable tool for diseases that are currently challenging or impossible to definitively delineate using current clinically available imaging tools, such as delineation of degenerative disk disease (sterile inflammation) from discitis osteomyelitis (infection). With the imaging tools now available, it is intriguing to envision a role in which imaging can rapidly diagnose infection ( 18 F-fluoromannitol: broad spectrum) and optimize the selection of an appropriate antibiotic for the pathogen ( 18 F-FDS: Enterobacterales specificity). Thus, these precision medicine tools may improve management of patient care and limit or eliminate unnecessary antibiotic use.
Our study was not without limitations. 18 F-fluoromannitol requires active transport of mannitol mediated by the mannitol-specific phosphotransferase system in bacteria. Thus, it is possible that senescent or slow-growing bacterial populations may diminish 18 F-fluoromannitol sensitivity. However, recent studies have shown that mannitol and fructose stimulated bacterial metabolism and enabled aminoglycoside antibiotic sensitivity (28,35,36). Further studies may be warranted to examine whether 18 F-fluoromannitol can serve as a prognostic indicator for this type of therapeutic strategy. Our studies revealed limited accumulation of 18 F-fluoromannitol in P. aeruginosa, a difficult-to-manage pathogen in patients with comorbidities (37,38). This outcome is surprising because the mannitol operon is well characterized in P. aeruginosa (39)(40)(41). Nonetheless, this finding is consistent with other mannitol-derived and sugar alcohol-derived radiopharmaceuticals studied (19) in P. aeruginosa in vitro. CONCLUSION We have described a novel radiopharmaceutical, 18 F-fluoromannitol, for imaging infections in a diverse spectrum of pathogenic organisms, including S. aureus, A. baumannii, and E. coli. Production of 18 F-fluoromannitol is straightforward, robust, and high-yielding, thus facilitating wide accessibility. Accordingly, 18 F-fluoromannitol might be rapidly translated to clinical studies as a noninvasive diagnostic tool facilitating rapid delineation of infection from sterile inflammatory processes, ultimately reducing the incidence of antimicrobial resistance promoted by selection pressures derived from unnecessary use of antibiotics. DISCLOSURE This work is supported by NIH/NIBIB R01EB028338-01 and the Military Infectious Diseases Research Program and Defense Health Programs subcontract DoD/NMRC N3239820P0034 (Kiel Neumann). Mark Simons and Alexandria Kesterson are military service members; this work was prepared as part of their official duties. Title 17, U.S. Code, §105, provides that copyright protection under this title is not available for any work of the U.S. government. Title 17, U.S. Code, §101, defines a U.S. government work as a work prepared by a military service member or employee of the U.S. government as part of that person's official duties. The views, opinions, or findings contained in this report are those of the authors and should not be construed as an official Department of the Navy, Department of Defense, or U.S. government position, policy, or decision unless so designated by other documentation. No other potential conflict of interest relevant to this article was reported.