Evaluating the Performance of Pathogen-Targeted Positron Emission Tomography Radiotracers in a Rat Model of Vertebral Discitis-Osteomyelitis

Abstract

Background

Vertebral discitis-osteomyelitis (VDO) is a devastating infection of the spine that is challenging to distinguish from noninfectious mimics using computed tomography and magnetic resonance imaging. We and others have developed novel metabolism-targeted positron emission tomography (PET) radiotracers for detecting living Staphylococcus aureus and other bacteria in vivo, but their head-to-head performance in a well-validated VDO animal model has not been reported.

Methods

We compared the performance of several PET radiotracers in a rat model of VDO. [¹¹C]PABA and [¹⁸F]FDS were assessed for their ability to distinguish S aureus, the most common non-tuberculous pathogen VDO, from Escherichia coli.

Results

In the rat S aureus VDO model, [¹¹C]PABA could detect as few as 10³ bacteria and exhibited the highest signal-to-background ratio, with a 20-fold increased signal in VDO compared to uninfected tissues. In a proof-of-concept experiment, detection of bacterial infection and discrimination between S aureus and E coli was possible using a combination of [¹¹C]PABA and [¹⁸F]FDS.

Conclusions

Our work reveals that several bacteria-targeted PET radiotracers had sufficient signal to background in a rat model of S aureus VDO to be potentially clinically useful. [¹¹C]PABA was the most promising tracer investigated and warrants further investigation in human VDO.

Vertebral discitis-osteomyelitis (VDO) is a common and potentially devastating infection, especially in patients who are immunocompromised, are intravenous drug users, or have other risk factors for hematogenous spread of infection [1]. The most common nontuberculous cause of VDO in patients is Staphylococcus aureus, a gram-positive bacterium whose management has been challenged by the emergence of multidrug-resistant strains such as methicillin-resistant S aureus (MRSA) [2]. Not only can uncontrolled infection lead to sepsis, but abscess formation and granulation tissue can occupy the epidural space, leading to compression of the spinal cord. The diagnosis of VDO is difficult in both the acute and chronic setting and may present as a indolent disease, especially in older patients [3]. The current diagnostic gold standard requires invasive tissue sampling, Gram staining, and a positive culture to definitively diagnose infection and to determine antibiotic susceptibilities to guide optimal therapy. However, given the location of the nerve roots and the lack of external landmarks, biopsies of the spine may be low yield [4] and even potentially dangerous for the patient. In the case of documented infection, institution of effective antibiotic therapy is associated with improved outcomes. In contrast, unnecessary antibiotic treatment has associated morbidity, including disruption of the microbiome and selection for antibiotic-resistant organisms [5, 6].

The symptoms of VDO can be nonspecific, with laboratory testing including white blood cell (WBC) count, C-reactive protein, and erythrocyte sedimentation rate frequently inconclusive [7]. Both computed tomography (CT) and magnetic resonance imaging (MRI) are used in the evaluation of VDO in addition to nuclear imaging techniques [8–11]. Current imaging techniques are limited in the accurate diagnosis of VDO, as CT and MRI both rely on the presence of nonspecific structural abnormalities that often occur late in the disease process and can mimic processes such as osteoradionecrosis, rheumatologic disease, or age-related degeneration [12, 13]. The evaluation of VDO, especially in chronic cases, would benefit greatly from an imaging approach that is highly specific to live bacteria, both for diagnosis and monitoring antibiotic response.

Although many infections may be diagnosed using MRI and CT, it is frequently difficult to distinguish bacterial infection from noninfectious entities. Several nuclear imaging tools have been applied to this problem, in particular ¹¹¹In WBC scanning, in which the patient's own immune cells are radiolabeled and single-photon emission computed tomography (SPECT) is performed [14], and [⁶⁷Ga]gallium citrate scanning [15, 16] targeting the transferrin receptor and potentially bacterial siderophores [17]. Recently, the investigation of cardiovascular and other infections has included full-body positron emission tomography (PET) using 2-deoxy-2-[¹⁸F]fluoroglucose ([¹⁸F]FDG), the radiotracer widely used in oncologic imaging [18]. Like cancer cells, activated immune cells have increased glucose uptake and thus may be detected using FDG-PET. However, lack of specificity—for example, distinguishing infection from tumor—is a major problem for both SPECT and PET approaches. Furthermore, these imaging techniques are dependent on host inflammatory response, which may be reduced or absent in individuals with a compromised immune system, for example patients with human immunodeficiency virus/AIDS, or undergoing chemotherapy and thus at higher risk for infection [19].

To overcome these obstacles to precise identification of VDO and other infections, there has been a pronounced interest in developing methods to detect bacteria in vivo during active infection. Several imaging strategies have targeted diverse bacterial-specific metabolic pathways to yield fluorescent bacteria or, more recently, for PET imaging [20]. These approaches have included radiolabeled antibiotics, sugars, sugar alcohols, antibodies, and peptides, with some of the most promising approaches including 2-deoxy-2-[¹⁸F]fluorosorbitol ([¹⁸F]FDS) showing dramatic data in infected patients [21, 22]. In our laboratory, we have focused on PET-compatible D-amino acids including D-[methyl-¹¹C]methionine ([¹¹C]D-met) [23, 24] and D-[3-¹¹C]alanine ([¹¹C]D-ala [25], as well as the folate-targeted radiotracer α-[¹¹C]para-aminobenzoic acid ([¹¹C]PABA) [26, 27]. Based on published in vitro studies using Staphylococcus aureus (the most common causative organism in VDO [7, 28], we studied a VDO rat model [25, 29] using several bacteria-targeted PET radiotracers (Figure 1). We performed micro-PET-CT (μPET-CT) on infected animals, to show how dual-modality imaging (PET-CT or PET-MRI) might be applied to patients to not only identify the presence of infection, but spatially localize it. In combination with an anatomic imaging method, bacteria-targeted PET could be used to detect S aureus in multiple locations, for example within infected heart valves, prosthetic joints, and intervertebral discs.

Figure 1.

Tracers employed in the rat vertebral discitis-osteomyelitis (VDO) model. The ¹¹C-labeled tracers [¹¹C]PABA, [¹¹C]D-ala, and [¹¹C]D-met were recently developed by our laboratories and have been shown to be sensitive to the strain of Staphylococcus aureus studied. Both [¹⁸F]FDG and the single-photon emission computed tomography tracer [⁶⁷Ga]gallium citrate have been used clinically for decades, and applied to VDO, with the PET correlate [⁶⁸Ga]gallium citrate applied more recently to infected patients. The Enterobacteriaceae-sensitive [¹⁸F]FDS has been studied in numerous preclinical models of infection and humans and is not significantly accumulated in S aureus.

MATERIALS AND METHODS

Radiotracer Syntheses

In all cases, PET radiotracers were synthesized as we have previously described. [¹¹C]PABA was synthesized from a commercially available Grignard precursor [26]. [¹¹C]D-ala was synthesized using symmetric alkylation of glycine-derived precursors via [¹¹C]methyl iodide, in the presence of a phase-transfer cinchonidinium catalyst [25]. [¹¹C]D-met was synthesized in-loop from a linear D-homocysteine precursor in >99% enantiomeric excess [24]. [⁶⁸Ga]gallium citrate was generated by eluting [⁶⁸Ga] from a Ge-68/Ga-68 Generator with trapping on a cartridge and subsequent elution via sodium citrate [25]. [¹⁸F]FDG was synthesized by the University of California, San Francisco (UCSF) cyclotron facility using standard procedures [25], while [¹⁸F]FDS was produced via 2-electron reduction of [¹⁸F]FDG by sodium borohydride [21]. All tracers used had radiochemical purities >99% and radiochemical yields/molar activities similar (within 50%) to mean values previously reported.

Bacterial Strains and Growth Conditions

Bioluminescent S aureus Xen 29 (the strain studied throughout this work unless otherwise indicated) bacteria were aerobically grown in lysogeny broth (LB) containing 100 μg/L of kanamycin to an optical density at 600 nm of 1.0, followed by centrifugation and resuspension of the bacteria in an equal volume of 1× phosphate-buffered saline (PBS). Xen 29 is derived from the ATCC 12600 S aureus strain and possesses a stable copy of the Photorhabdus luminescens lux operon. Culture and preparation of Escherichia coli (ATCC 25922) was via an identical method, and heat-killed bacteria for all experiments were obtained using previously reported techniques [25].

Rat VDO Model

All animal procedures were approved by the UCSF Institutional Animal Care and Use Committee, and all studies were performed in accordance with UCSF guidelines regarding animal housing, pain management, and euthanasia. Sprague-Dawley rats (male, 10–12 weeks old, Charles River Laboratories) were used for all experiments. Rats were inoculated in the third intervertebral space with 50 μL of PBS containing 1.4 × 10⁷ colony-forming units (CFU) live bacteria or a 10-fold higher dose of heat-killed bacteria [21] into the third/sixth intervertebral spaces from the base of the tail at 50% depth (based on the diameter of the tail) as described previously [29]. The rats were imaged at 2, 4, 6, 8, and 10 days using a Xenogen IVIS 50 instrument or Inveon μPET-CT (Siemens, Erlangen, Germany) following injection with a PET radiotracer.

Radiotracer Screening

Several cohorts (n = 5 for [¹¹C]PABA; n = 5 for [¹¹C]D-ala; n = 3 for the remaining tracers) were inoculated with S aureus in the third intervertebral space, and 10× heat-killed S aureus in the fifth intervertebral space. Intervertebral infection was confirmed by Xenogen IVIS 50, followed by μPET-CT at 4 days. Region of interest (ROI) analysis from PET images was used to compare tracer performance.

[¹¹C]PABA Dynamic Imaging

Three rats were inoculated with S aureus in the third intervertebral space as above, with intervertebral infection confirmed by Xenogen IVIS 50 at 2 days. This cohort was subsequently studied via [¹¹C]PABA μPET-CT at 2, 4, 6, and 10 days via ROI analyses as above.

Infection Burden Analysis

Rats (n = 3 per group) were inoculated with 50 μL PBS containing 10⁶, 10⁴, or 10² CFU S aureus in the third intervertebral space. The inoculum was confirmed by serial dilution and plating as described previously [25]. The animals were imaged by Xenogen IVIS 50 and [¹¹C]PABA μPET-CT at 4 days, and then euthanized. Spinal sections were isolated and counted using gamma counting (Hidex, Turku, Finland), followed by homogenization of the spinal sections in a small volume of 1× PBS. Serial dilutions of harvested tissues homogenates were plated on LB 1% agar plates to quantify bacterial burden.

Dual Imaging of S aureus and E coli–Infected VDO

Three rats were inoculated with 1.4 × 10⁷ CFU E coli in the third intervertebral space and 1.4 × 10⁷ CFU S aureus in the fifth intervertebral space. They were subsequently imaged using Xenogen IVIS 50 and a dual tracer imaging protocol at 4 days, whereby the rats were administered [¹¹C]PABA (t = 0), imaged by μPET-CT (t = 50 minutes), administered [¹⁸F]FDS (t = 80 minutes), and imaged by μPET-CT (t = 200 minutes). ROI analysis from PET images was used to show uptake at the 2 injection sites by E coli and S aureus.

Imaging

Bioluminescence

Animals were imaged on a Xenogen IVIS 50 to confirm technically successful intervertebral inoculation of S aureus. In vivo bioluminescence imaging data are presented on a color-scale overlaid on a grayscale photograph of mice and displayed as radiance (photons/second/cm²/steradian) within a circular ROI. Images were analyzed using Living Image software (PerkinElmer). Bioluminescence imaging was primarily used to establish technically successful inoculation for subsequent PET imaging, without explicit analysis or correlation to PET tracer uptake.

μPET-CT

The same general protocol was used for all studies. A tail vein catheter was placed in rats under isoflurane anesthesia. For carbon-11 studies, approximately 800 μCi of radiotracer was injected via the tail vein catheter. For [⁶⁸Ga]gallium citrate studies, approximately 250 μCi of [⁶⁸Ga]gallium citrate was injected via the tail vein catheter. For fluorine-18 studies, approximately 150 μCi of radiotracer was injected via the tail vein catheter. The animals were placed on a heating pad to minimize shivering. Mice were allowed to recover and micturate, and at the time points indicated (40 minutes for carbon-11 tracers, 120 minutes for fluorine-18 tracers, and 50 minutes/200 minutes for Gram staining), were placed back under isoflurane anesthesia. The animals were then transferred to the μPET-CT system and imaged using a single static 20-minute PET acquisition followed by a 10-minute micro-CT scan for attenuation correction and anatomical co-registration. No adverse events were observed during or after injection of any compound. Anesthesia was maintained during imaging using isoflurane. As described in the Results section, several cohorts were studied using ex vivo biodistribution analysis following completion of imaging and sacrifice. Gamma counting of harvested tissues was performed using a Hidex Automatic Gamma Counter as above.

Data Analysis and Statistical Methods

Both ROI and ex vivo biodistribution analyses were performed.

ROI Analysis

μPET-CT data were analyzed using the open source software AMIDE [30] and percent injected dose per cubic centimeter (%ID/cc) was used for quantitative comparison. %ID/cc values were established via 8 mm³ ROIs using the elliptical tool.

Ex Vivo Analysis

Following some studies, the rats were killed and the spinal tissue was homogenized for Hidex gamma counting studies, and serial dilution (1× PBS) and plating to quantify the CFU present at the time of imaging.

Data Representation

All statistical analyses were performed using Prism software version 9.0 (GraphPad, San Diego, California). Live versus heat-killed comparisons were analyzed using paired t tests. Dynamic imaging data were analyzed using repeated-measures 1-way analysis of variance, followed by Tukey multiple comparison tests. Dose responses of [¹¹C]PABA uptake were analyzed using unpaired t tests. P < .05 was considered statistically significant. All graphs are depicted with error bars corresponding to the standard error of the mean.

Please see the Supplementary Materials and Methods for detailed information regarding microbiology, radiosynthesis, and several in vitro studies.

RESULTS

Bioluminescent Imaging of S aureus Xen 29-Infected Rats Reveal the Highest In Vivo Signals at 4 Days Postinfection in the Rat VDO Model

In this study, we used a previously published rat VDO model [29], in which we have previously examined the behavior of the bacterial metabolism-specific tracer [¹¹C]D-ala. The goal of using bioluminescent S aureus was to establish technically successful intervertebral inoculation for subsequent analysis using μPET-CT. Using a Xenogen IVIS 50 instrument, we performed a time course of S aureus infection using a strain transformed with a bioluminescent plasmid, Xen 29, by inoculating the third intervertebral space. As qualitative signals from bioluminescent S aureus were maximal at day 4 (Figure 2), we used this time point for subsequent imaging studies using PET. The subsequent loss of optical signal over time has been reported for bioluminescent S aureus in this model, which reflects proliferation of the wild-type S aureus over time in the absence of plasmid maintenance in vivo [29].

Figure 2.

Optical imaging of the rat vertebral discitis-osteomyelitis model using bioluminescent Staphylococcus aureus. A, S aureus Xen 29 was inoculated at the third intervertebral space followed by serial imaging on a Xenogen IVIS 50 instrument as described in the Materials and Methods. Shown are representative bioluminescent signals at the site of inoculation on days 2, 4, and 6, indicative of technically successful intervertebral inoculation. B, Quantitative analysis of bioluminescent imaging performed at 2, 4, 6, 8, and 10 days demonstrates that day 4 is the optimal time for bioluminescent imaging of positron emission tomography tracers.

Performance of [¹¹C]PABA, [¹¹C]D-Ala, [¹¹C]D-Met, and [¹⁸F]FDS in the Rat S aureus VDO Model as Assessed by PET

This study used several PET radiotracers targeting bacterial metabolism developed recently in our laboratories ([¹¹C]PABA, [¹¹C]D-ala, [¹¹C]D-met, [¹⁸F]FDS) (Figure 1) and compared them to the established tracers [⁶⁸Ga]gallium citrate and [¹⁸F]FDG. Rats were infected with live or 10-fold excess heat-killed S aureus in the third intervertebral disc space [25]. Following IVIS confirmation of infection, we performed μPET-CT using these 6 tracers and analyzed the data using 8-mm spherical ROIs on AMIDE [30] for quantitative comparison. The visual differences between animal groups were striking (Figure 3A), with significantly higher uptake for S aureus compared to a 10-fold excess of heat-killed S aureus for [¹¹C]PABA, [¹¹C]D-ala, [¹¹C]D-met, and [⁶⁸Ga]gallium citrate (Figure 3B; P = .0001, .0001, .0008, and .0098, respectively). In contrast, uptake of [¹⁸F]FDG was similar for live and heat-killed S aureus (P = .5900), and there was no discernable uptake of [¹⁸F]FDS by either live or heat-killed S aureus. The corresponding fold differences (Figure 3C) show that [¹¹C]PABA had the highest signal-to-noise ratio (ie, live/heat-killed ratio) in this model, approximately 23-fold.

Figure 3.

Performance of positron emission tomography (PET) tracers in the rat Staphylococcus aureus vertebral discitis-osteomyelitis (VDO) model at 4 days postinoculation. Rats were inoculated with S aureus (live) or a 10-fold excess of heat-killed (HK) bacteria in adjacent intervertebral spaces on day 0 with technically successful VDO infections confirmed using a Xenogen IVIS 50 instrument and studied using micro-positron emission tomography–computed tomography (μPET-CT) on day 4. A, μPET-CT imaging of infected rats (n = 5 for [¹¹C]PABA, n = 5 for [¹¹C]D-ala N = 3 for other tracers). PET signals at day 4 postinoculation were detected at the site of inoculation for all tracers except [¹⁸F]FDS. B, Quantitative region of interest (ROI) analyses comparing inoculation of S aureus, 10-fold excess of HK S aureus, or background for the indicated tracers (background ROI was plotted for reference only). Statistically significant differences between live and HK levels were observed for [¹¹C]PABA (P = .0003, n = 5), [¹¹C]D-ala (P = .0011, n = 4), [¹¹C]D-met (P = .0092, n = 3), and [⁶⁸Ga]gallium citrate (P = .0011, n = 3), but not for [¹⁸F]FDG (P = .47, n = 3). No [¹⁸F]FDS uptake was observed for either live or HK S aureus. C, Ratio of live vs 10-fold excess of HK S aureus derived from ROI analyses of PET scans. Abbreviations: BG, background; HK, heat-killed; ID/cc, percent injected dose per cubic centimeter.

Dynamic Imaging of Rat VDO Model Using [¹¹C]PABA-PET Shows Increasing Signals Over Time

We next performed a time course of [¹¹C]PABA in 3 rats per cohort inoculated in the third intervertebral disc space with S aureus. Serial μPET-CT revealed increasing PET signals over the period of 10 days (P = .28), as shown in Figure 4. In contrast, the bioluminescent signals decreased over time in this model, as reported previously [29]. ROI analysis (%ID/cc for comparison) showed that relative to day 2, [¹¹C]PABA had approximately 2.2-fold higher uptake on day 4 (adjusted P = .015), approximately 2.5-fold higher uptake on day 6 (adjusted P = .086), and approximately 2.9-fold higher uptake on day 10 (P = .028).

Figure 4.

Dynamic study of [¹¹C]PABA in the rat vertebral discitis-osteomyelitis model. Mice were injected with [¹¹C]PABA on days 2, 4, 6, and 10 (n = 3 per cohort) after Staphylococcus aureus vertebral inoculation. A, Representative rat micro-positron emission tomography (PET)–computed tomography studies on days 2, 6, and 10. B, Graphical depiction of increasing PET signals over time. Abbreviation: ID/cc, percent injected dose per cubic centimeter.

[¹¹C]PABA Uptake Is a Sensitive Probe of S aureus Infection in the Rat VDO Model

We investigated the limit of detection of S aureus via intervertebral inoculation decreasing inocula: 10⁶, 10⁴, or 10² CFU were inoculated, and the bacterial burden at the time of imaging was determined by homogenization, serial dilution and plating, with CFUs rounded to the nearest order of magnitude. We observed an approximately 10-fold increase for each of the inocula (10⁷ 10⁵, and 10³, respectively). μPET-CT imaging showed detectable [¹¹C]PABA accumulation at 3 doses (Figure 5B), which was further confirmed by ex vivo gamma counting (Supplementary Figure 1). Compared to PBS-inoculated segments, ROI analysis showed approximately 28-fold higher uptake (P = .0001) for 10⁷ CFU, approximately 23-fold higher uptake (P = .0001) for 10⁵ CFU, and approximately 6-fold higher uptake (P = .0016) for 10³ CFU.

Figure 5.

Dose response of [¹¹C]PABA uptake by Staphylococcus aureus in the rat vertebral discitis-osteomyelitis model. A, Cohorts of rats (n = 3 per group) were inoculated with either 10², 10⁴, or 10⁶ colony-forming units (CFU) S aureus and studied on day 4 using micro-positron emission tomography–computed tomography. The bacterial burdens were determined by homogenization of the spinal sections. B, Percent injeted dose per cubic centimeter (ID/cc) for each of the different inoculums and bacterial burdens. In all cases, infected intervertebral discs were compared to control (noninoculated rats) by region of interest analysis (10⁷, ∼28-fold, P = .0001; 10⁵ ∼23-fold, P = .0001; 10³ ∼6-fold, P = .0016).

Simultaneous Detection of Gram-Positive and Gram-Negative Bacteria Using [¹¹C]PABA and [¹¹C]FDS

[¹⁸F]FDS has been shown to be avidly and specifically taken up in vivo by gram-negative bacteria, with little or no uptake seen in gram-positive bacteria [21]. While this limits the utility of ¹⁸F-FDS as the sole imaging tracer in the diagnosis of VDO, it does suggest the possibility of an in vivo “Gram stain” equivalent by simultaneously imaging with [¹¹C]PABA to detect both common gram-positive and gram-negative bacteria associated with VDO and with ¹⁸F-FDS to detect gram-negative bacteria associated with VDO. To test this strategy, we employed the rat VDO model in which we inoculated S aureus into the third intervertebral space and E coli ATCC 25922 into the fifth intervertebral space. The animals were injected sequentially with 800 μCi of [¹¹C]PABA and imaged at t = 50 minutes, followed by administration of 150 μCi of [¹⁸F]FDS at t = 80 minutes and imaging at 200 minutes (allowing carbon-11 decay). The total delay between the injection of [¹¹C]PABA and the initiation of [¹⁸F]FDS imaging was approximately ten carbon-11 half-lives, making the radiotracer signal from the first scan noncontributory to the second scan. Subsequently, ROI analyses of the infected areas were performed for both time points. As shown in Figure 6, in S aureus infection, [¹¹C]PABA was retained but not [¹⁸F]FDS. However, in an E coli–infected cohort, PET signals were detected using both [¹¹C]PABA and [¹⁸F]FDS. These differences could be used to define both the location and to identify 2 different bacteria, at least for the E coli and S aureus strains used in this study.

Figure 6.

Comparison of [¹¹C]PABA and [¹⁸F]FDS of rat vertebral discitis-osteomyelitis model of Staphylococcus aureus or Escherichia coli by micro-positron emission tomography–computed tomography (μPET-CT). A, [¹¹C]PABA and [¹⁸F]FDS was serially administered to rats inoculated with both S aureus and E coli at different intervertebral segments on day 4 (n = 3), with μPET-CT subsequently performed. Both S aureus and E coli infections can be detected using [¹¹C]PABA, whereas only E coli can be detected using [¹⁸F]FDS. B, Graphical depiction of data. C, Timeline for tracer injection and imaging time-points. Abbreviations: EC, Escherichia coli; ID/cc, percent injected dose per cubic centimeter; SA, Staphylococcus aureus.

DISCUSSION

Numerous PET and SPECT radiotracers have been used to study S aureus in vivo, including the clinical tracers [¹⁸F]FDG and [⁶⁷Ga]gallium citrate in a rat VDO model [15, 31, 32]. More recently, several elegant approaches have been developed to improve the specificity of live S aureus detection. These include the use of sugars including 6′′-[¹⁸F]-fluoromaltotriose [33, 34], which targets the bacterial maltose transport mechanism, and probes incorporated into the folate biosynthesis pathway, such as [¹⁸F]fluoropropyl-trimethoprim [35] and [¹⁸F]-labeled PABA derivatives [36]. Bacterial nitroreductase has been recently targeted for S aureus imaging, via both [¹⁸F]PABA derivatives (using a nitro-prodrug) [37] and nitrogen mustard analogues [38]. Other diverse approaches use [⁶⁸Ga]-labeled bacterial siderophores (ie, desferrioxamine-B) [39], the host protein sphingosine-1-phosphate receptor 1 [40], the 6-position phosphorylated analogue of FDG [41], and the [^99mTc]-labeled antimicrobial peptide ubiquicidin [42]. Some reports have also compared the performance of multiple tracers in preclinical models of S aureus infection [43, 44].

In our study, several PET tracers targeting bacterial metabolism ([¹¹C]PABA, [¹¹C]D-ala, [¹¹C]D-met) showed significant uptake compared to heat-killed bacteria in S aureus spinal infection, with [¹¹C]PABA showing the most promise given the high infection/background (>20-fold) ratio. We performed additional studies to suggest how the variable sensitivities of PET tracers might be leveraged in future clinical applications, for instance, in distinguishing gram-negative from gram-positive infections. For example, the more broadly sensitive [¹¹C]PABA could potentially be used in conjunction with the Enterobacteriaceae-targeted [¹⁸F]FDS to identify broad classes of pathogens in vivo. We demonstrated the feasibility of this concept using S aureus Xen 29 and E coli ATCC 25922 in a small number of animals, but to be sufficiently robust in clinical practice, numerous additional data are needed. First, other bacterial causes of VDO need to be included in the analysis, for example, Streptococcus pneumoniae, Enterococcus faecalis, Mycobacterium tuberculosis, and Salmonella species. The PET tracer uptake of Staphylococcus epidermidis and other coagulase-negative staphylococci would be particularly important, since these (frequently commensal) organisms are often discovered in biopsy samples, confounding clinical management [45]. Finally, we and others have observed marked variability in PET tracer accumulation by multiple strains of the same species, for example, 2-deoxy-2-[¹⁸F]fluoromannitol uptake by different S aureus clinical isolates [46]. Therefore, additional clinical strains of S aureus (including MRSA) and E coli should be investigated in the context of the described VDO model. Future preclinical analyses will include additional organisms and strains to better identify potential causes of false-negative and false-positive patient PET exams.

In addition to the incomplete set of pathogens studied, a major limitation of this study was the small number of animals used, necessitated by the rat VDO model, preclinical μPET-CT, and cyclotron use. Furthermore, we should emphasize that we sought a radiotracer that was sensitive to, rather than specific for, S aureus. The described approach cannot positively identify S aureus as a causative organism in vivo since numerous pathogens will accumulate [¹¹C]PABA or [¹¹C]D-ala. We do not have the techniques to identify individual species, and certainly not strains of a given species in vivo. In the future, these tools may be afforded by highly unique bacterial components (ie, N-acetyl muramic acid–related products in certain Pseudomonas aeruginosa strains [47]) or siderophore-mediated imaging strategies [48]. Finally, from a radiochemical standpoint, the use of the short t_1/2 [¹¹C]radionuclide presents a major obstacle to synthesizing and distributing PET tracers beyond the outpatient, academic setting. The discovery of new [¹⁸F]-labeled tracers accumulated by S aureus is therefore critical for high clinical impact.

CONCLUSIONS

The goal of this study was to find a pathogen-targeted PET tracer that was sensitive to S aureus infection in a murine VDO model. Several PET tracers targeting bacterial metabolism ([¹¹C]PABA, [¹¹C]D-ala, [¹¹C]D-met) showed significant uptake in S aureus spinal infection, with [¹¹C]PABA showing special promise given the high infection/background (>20-fold) ratio. The more broadly sensitive [¹¹C]PABA could potentially be used in conjunction with the Enterobacteriaceae-targeted [¹⁸F]FDS to distinguish gram-positive from gram-negative infection in vivo.

Supplementary Data

Supplementary materials are available at The Journal of Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.

Notes

Author contributions . M. F. L. P., D. M. W., M. A. O., and S. K. J. proposed and supervised the overall project. M. F. L. P., J. B., H. Q., A. M. S., S. J. R., and S. L. performed or supported the radiochemistry. M. F. L. P., A. A. A., and J. M. L. developed the cell cultures for in vitro studies. M. F. L. P., A. M. S., H. Q., M. L.-A., P. A. P.-G., and I. P. performed the μPET-CT imaging studies, and M. F. L. P. and O. S. R. performed subsequent data analysis. M. F. L. P. and M. L.-A. performed ex vivo analysis. M. F. L. P., D. M. W., S. K. J., A. A. O., R. R. F., M. L.-A., H. Q., and J. E. wrote and edited the manuscript.

Financial support. This work was supported by the National Institutes of Health (grant numbers R01 EB024014, R01 EB025985, and R01 EB030897) and the Department of Defense (grant number DOD A132172).

Supplement sponsorship. This article appears as part of the supplement “Seeing Is Believing: The Potential of Molecular Imaging Approaches for Infectious Diseases,” sponsored by the Center for Infectious Disease Imaging, Clinical Center, NIH (Intramural Research Program), and the Center for Infection and Inflammation Imaging Research, Johns Hopkins University School of Medicine, with individual contributions from Long Island Jewish Medical Center; University of California, San Francisco; and University of Maryland School of Pharmacy.