Imaging sensitive and drug-resistant bacterial infection with [11C]-trimethoprim

Abstract

BACKGROUNDSeveral molecular imaging strategies can identify bacterial infections in humans. PET affords the potential for sensitive infection detection deep within the body. Among PET-based approaches, antibiotic-based radiotracers, which often target key bacterial-specific enzymes, have considerable promise. One question for antibiotic radiotracers is whether antimicrobial resistance (AMR) reduces specific accumulation within bacteria, diminishing the predictive value of the diagnostic test.METHODSUsing a PET radiotracer based on the antibiotic trimethoprim (TMP), [11C]-TMP, we performed in vitro uptake studies in susceptible and drug-resistant bacterial strains and whole-genome sequencing (WGS) in selected strains to identify TMP resistance mechanisms. Next, we queried the NCBI database of annotated bacterial genomes for WT and resistant dihydrofolate reductase (DHFR) genes. Finally, we initiated a first-in-human protocol of [11C]-TMP in patients infected with both TMP-sensitive and TMP-resistant organisms to demonstrate the clinical feasibility of the tool.RESULTSWe observed robust [11C]-TMP uptake in our panel of TMP-sensitive and -resistant bacteria, noting relatively variable and decreased uptake in a few strains of P. aeruginosa and E. coli. WGS showed that the vast majority of clinically relevant bacteria harbor a WT copy of DHFR, targetable by [11C]-TMP, and that despite the AMR, these strains should be "imageable." Clinical imaging of patients with [11C]-TMP demonstrated focal radiotracer uptake in areas of infectious lesions.CONCLUSIONThis work highlights an approach to imaging bacterial infection in patients, which could affect our understanding of bacterial pathogenesis as well as our ability to better diagnose infections and monitor response to therapy.TRIAL REGISTRATIONClinicalTrials.gov NCT03424525.FUNDINGInstitute for Translational Medicine and Therapeutics, Burroughs Wellcome Fund, NIH Office of the Director Early Independence Award (DP5-OD26386), and University of Pennsylvania NIH T32 Radiology Research Training Grant (5T32EB004311-12).

Keywords: Bacterial infections; Diagnostic imaging; Infectious disease.

Figures

Figure 1. Structure of [11C]-TMP and in vitro TMP dose-response assays of different bacterial strains.

(A) Structures of trimethoprim (TMP) and [11C]-TMP. (B) TMP dose-response assay on bacterial strains. OD600 measurement was taken following a 6-hour incubation of different bacterial strains with TMP. The susceptibility or resistance of a bacterial strain to TMP is color-coded based on the IC50 and minimum inhibitory concentration (MIC). Blue indicates susceptible bacteria, and red indicates resistant bacteria. n = 3; data represent mean ± SD. (C) Representative [11C]-TMP uptake in bacterial cultures after a 30-minute incubation at 37°C. n = 3–5; data represent mean ± SD. The experiment was repeated a total of 2–3 times for biological replicates.

Figure 2. Bioinformatic analysis of clinically relevant bacteria.

(A) Proportion of clinically relevant bacterial strains from the NCBI RefSeq database, with the indicated number of DHFR genes per genome. (B) Resistance characterization of DHFR genes in relevant bacterial strains from the NCBI RefSeq database. Each dot represents a strain.

Figure 3. CONSORT diagram of the [11C]-TMP study.

*Laboratory test results may have been collected from the medical record if they were completed within 30 days of screening; in these cases, they were not repeated for the purposes of this study. Refusal of labs did not preclude a patient from the study. **Surgical or systemic therapy was started if clinically indicated at the judgment of treating physicians. ***For patients who received systemic antibiotic therapy, this may have been within 1 week of therapy cessation. For patients who received surgical management, this may have occurred within 3–6 weeks after surgery, as clinical appropriate. Given that some patients are on chronic antibiotics, this scan may have occurred after completion of alternative or more intensive antibiotic therapy. $In some cases, and at the discretion of the investigator, the patient may have been scanned while on i.v. antibiotics.

Figure 4. Biodistribution of [18F]-FDG versus [11C]-TMP in a patient with lung cancer and biodistribution in a patient with underlying chronic lung disease.

(A) The image on the left shows a 64-year-old man with known lung adenocarcinoma who underwent a [18F]-FDG (549 MBq) and then a [11C]-TMP (563 MBq) PET/CT 2 days later. The [18F]-FDG image was acquired starting 71 minutes after injection. Whole-body maximum intensity projection (MIP) images demonstrate the difference in biodistribution of the tracers. In the lungs, [18F]-FDG is taken up both by metabolically active tumor and inflammatory cells, whereas [11C]-TMP is not. The image on the right shows a comparison MIP image of a patient with cystic fibrosis and chronic lung infections who underwent a [11C]-TMP PET/CT (780 MBq). The image was acquired starting 78 minutes after injection. PET images are scaled at 0–7 g/mL SUV. (B) [11C]-TMP imaging of a 44-year-old woman with several foci of infection in the chest (red arrows). Other sites of signal include the liver, the kidneys, red bone marrow, and the stomach. PET images are scaled at 0–7 g/mL SUV.

Figure 5. Time activity curves and bacterial heterogeneity.

(A and B) [11C]-TMP PET/CT images of a 44-year-old woman with an acute exacerbation of cystic fibrosis (A) before and (B) after treatment with 2 weeks of i.v. ceftriaxone. Regions of interest were drawn around 2 separate pulmonary airspace opacities. In addition, reference regions were also drawn around a reference lymph node and paraspinal musculature and within the aorta. Comparing PET/CT images before and after treatment, the visible changes in relative [11C]-TMP uptake in lesion 1 compared with lesion 2 demonstrate the bacterial heterogeneity and an apparent new infection with S. aureus based on sputum cultures. The patient received 487 MBq and 780 MBq of [11C]-TMP at the first and second time points, respectively. PET images are scaled at 0–5 g/mL SUV and CT images are scaled at –1,024 to +300 HU.

Figure 6. Acute exacerbation of cystic fibrosis.

(A) A 64-year-old man with known lung adenocarcinoma underwent a [18F]-FDG (549 MBq) and then a [11C]-TMP (563 MBq) PET/CT 2 days later. The [18F]-FDG image was acquired starting 71 minutes after injections. Whole-body maximum intensity projection (MIP) images demonstrate the difference in biodistribution of the tracers. In the lungs, [18F]-FDG is taken up both by metabolically active tumor and inflammatory cells, whereas [11C]-TMP is not. (B) A comparison MIP image of a 44-year-old woman with cystic fibrosis and chronic lung infections, who underwent a [11C]-TMP PET/CT (780 MBq). The image was acquired starting 78 minutes after injection. The PET images show several foci of infection in the chest (red arrows). Other sites of signal include the liver, the kidneys, red bone marrow, and the stomach. PET images are scalled 0-7 g/mL SUV.

Figure 7. Biopsy proven discitis osteomyelitis treated with antibiotics.

A 55-year-old man with clinically suspected lumbar discitis osteomyelitis was scanned with [11C]-TMP PET/CT at the initiation of empiric antibiotic therapy and after 6 more weeks of targeted antibiotic therapy. (A) Time line of patient disease sequelae and imaging. (B) Axial PET/CT images show a clear site of asymmetric [11C]-TMP uptake in the left L4–L5 facet at the start of therapy and lack of uptake after 6 weeks of i.v. treatment. Facet biopsy of the left L4–L5 facet grew methicillin-sensitive S. aureus. Note: the patient received different doses of [11C]-TMP, 129 MBq at the first time point and 672 MBq at the second time point; thus, the image quality was noisier at the first time point. PET images are scaled at 0–5 SUVmax. (C) The temporal morphologic sequelae of discitis osteomyelitis are demonstrated by sagittal CT images before, during, and after treatment. (D) In contrast to the PET/CT images, the gadolinium-enhanced MRI images of the patient at 10 weeks after therapy continue to demonstrate marrow replacement and contrast enhancement, findings that are nonspecific for active infection versus continued inflammation.