Impact of PET/CT among patients with suspected mycotic aortic aneurysms

Purpose To determine the impact of 18F-fluorodeoxyglucose (FDG) positron emission tomography/computed tomography (PET/CT) on clinical management in patients with suspected mycotic aortic aneurysms (MAA). Materials and methods For this observational cohort study 101 PET/CT were acquired in 50 patients, thereof 50 for the initial diagnosis/baseline scan, 51 for follow-up. Impact on patient management was defined in three categories: PET/CT results were “confirmed” (by clinical follow-up), “suspected” (conclusive, not confirmed), or “misleading” (proven wrong by follow-up). For clinical follow-up patient data were recorded at the time of imaging, and at the latest recorded clinical visit. It included patient demographics, clinical information, laboratory data, results of microbiology and other diagnostic procedures, information about treatment, and patient’s general health condition. Results In four patients (8%) no clinical follow-up was feasible, the other 46 patients were clinically followed for a median of 898 days (IQR 320–4105). The combined evaluation of all 101 PET/CT demonstrated an impact on patient management in 78,5% of cases (48,5% confirmed, 30% suspected). Results of 21,5% of the PET/CT examinations were misleading. Respective values at baseline and at follow-up were: impact on patient management in 82% and 74,5% (70% and 27.5% confirmed, and 12% and 47% suspected), misleading cases in 18% and 25.5%. Conclusion In MAA, PET/CT has a high impact on patient management, which is more pronounced with baseline than with follow-up examinations. However, PET/CT results may be misleading in a smaller proportion of cases.


Introduction
The impact of 18 F-fluorodeoxyglucose (FDG) positron emission tomography/computed tomography (PET/CT) on clinical management of cancer patients based on a large prospective data registry has recently shown that physicians change their intended management in 37% of cases after PET/CT [1]. Comparable data for the impact of PET/CT in infectious diseases is scarce. For example, Leroy-Freschini et al. [2] described an impact of PET/CT in the management of immunocompromised patients with invasive fungal infections in 55% of their patients at initial staging and in 46% at follow-up. Moreover, the detection of peripheral emboli with PET/CT in patients with endocarditis or cardiac device infections induced therapeutic modifications in 24-44% of patients [3]. For mycotic aortic aneurysms (MAA), the actual impact of PET/CT on patient management has not been evaluated, yet.
MAA are infectious arterial aneurysms, caused by microbial inoculation of aortic endothelium during bacteraemia [4], and which account for 0.7-4.5% of all aortic aneurysms [5]. Surgical and medical treatment of mycotic aortic aneurysms (MAA) is demanding. Evidence indicates that endovascular aortic repair (EVAR) of MAA may be an equivalent treatment option to open surgical treatment [6]. However, EVAR of MAA inevitably leads to secondary vascular graft infections, requiring long-term antimicrobial therapy.
Regarding imaging of MAA, recent publications have demonstrated a high diagnostic accuracy of PET/CT in the detection of MAA [7], and the useful of PET/CT in therapy control of MAA [8]. Hence, the aim of the present study was to determine the impact of PET/CT on clinical management of patients with suspected MAA.

Study design and definitions
Two different patient groups were included: i) prospective patients aged 18 years or older with suspected MAA and open and/or endovascular surgery enrolled in the Vascular Graft Cohort Study (VASGRA), and ii) retrospective patients with suspected MAA, who were examined with PET/CT between the years 2005 and 2018. The study was approved by the local ethics committee, namely the Kantonale Ethikkomission Zürich (protocol number 2018-01904), and we obtained written informed consent from all participants who were either prospectively enrolled or examined between the years 2016 and 2018; for subjects scanned between the years 2005 and 2015, written informed consent was waived due to retrospective inclusion by the local ethics committee, namely the Kantonale Ethikkomission Zürich (protocol number 2018-01904). All procedures were in accordance with the 1964 Helsinki declaration and its later amendments.
Patient demographics are described in Table 1 and S1 Table. Consecutively and prospectively enrolled patients had PET/CT scans at baseline, during follow-up on antimicrobial therapy, and at the end of antimicrobial treatment. In some patients, another control PET/CT three months after the end of antimicrobial therapy was acquired. For all retrospectively enrolled patients, PET/CT was acquired depending on the clinical condition of the patient. A chart review at each time of imaging, and at the last recorded clinical visit was performed. It included patient demographics, clinical information, laboratory and microbiological data, results from other diagnostic procedures, and information about treatment.
According to the MAGIC criteria [9] a secondary vascular graft infection was diagnosed if a stentgraft was placed in an infected area. In case of a native infected vessel, MAA was diagnosed in an overall appraisal of clinical presentation, laboratory and imaging [6,10].

Imaging data acquisition
Five different types of PET/CT scanners were used in our institution during the 15-year study period (i.e., Discovery ST16, Discovery VCT, Discovery MI, Discovery 690 and Discovery 710, all GE Healthcare, Waukesha, WI). All examinations were performed with the following imaging standards: a) PET/CT scan range at least from the vertex of the skull to the mid-thighs, b) non-enhanced CT scans for attenuation correction, c) all patients fasted for at least 4 hours and had no insulin injections within four hours before to the FDG administration, d) bodyweight adapted intravenous injection of FDG after supine resting for a standardized uptake time of 60 minutes.
Additional contrast-enhanced CT of the chest and/or abdomen was performed as part of the PET/CT examination in a subset of patients, using a tube voltage of 120 kV and a tube current-time product of 100-320 mAs. CT was timed for imaging at arterial phase and portal venous phase after intravenous injection of 80 ml of iodinated contrast medium (Visipaque 1 320, GE Healthcare, Waukesha, WI).
All imaging data sets were analysed independently by two double board-certified radiologists and nuclear medicine physicians with 12 and 13 years of experience in hybrid imaging on a AW workstation version 4.7 (GE Healthcare Biosciences, Pittsburgh, PA). Primary and secondary diagnoses (e.g., infectious foci not in the vicinity of the aorta or other relevant findings such as malignancy) were documented. A consensus reading was performed if results differed.

Impact on patient management
Impact on patient management was defined by three categories: "confirmed", "suspected" and "misleading". These were defined as follows: Confirmed impact was noted if the PET/CT report conclusively answered the referral question or detected a relevant additional incidental finding, which was previously unknown to the

PLOS ONE
referring physicians and changed patient management. All PET/CT findings, categorized as "confirmed", had to be confirmed by clinical follow-up and/or other diagnostic modalities (e.g. PET-positive MAA confirmed by microbiology, or incidentally diagnosed cancer confirmed by histology).
Suspected impact was noted, if the PET/CT report either clearly answered the referral question or described an unsuspected and presumably relevant finding, however, the "suspected" impact could not be confirmed by other diagnostic modalities or by follow-up.
A misleading impact was noted, if the PET/CT report either clearly answered the referral question or described an incidental finding, but the findings were proven wrong either by follow-up or other diagnostic modalities.
Notably, if more than one referral question was to be answered by PET/CT, impact on management was noted if the main question was answered. In detail, if the referral question was just "infectious aneurysm?", impact was noted, if PET/CT correctly confirmed or correctly ruled out the suspicion of MAA. However, if the question was "infectious aneurysm or other infectious foci?", impact was only noted if PET/CT correctly described a particular site of infection; if PET/CT did not identify a particular site of infection, no impact on management was noted. The latter was the case, because the patient clinically had an infection, and hence the PET/CT scan must have been considered false negative. Furthermore, if the referral question was just "follow-up" of a known site of infection, impact on management was noted, if PET/CT results (e.g. decrease of FDG-activity) elicited a change in patient treatment (e.g., termination of antimicrobial treatment).

Statistical analyses
Statistical analysis was performed using commercially available software (Stata, Version 16, StataCorp, College Station, Texas). Variables were expressed as median and IQR (25th, 75th percentiles) or percentages. One-sided Fisher's exact tests were performed to determine differences in impact on patient management with regard to the referral question for the initial PET/CT. A P-value of < .05 was considered to indicate statistical significance, and a P-value between.05 and.1 was considered to indicate a statistical trend.

Patient population
Six patients were excluded because no written consent was signed (Fig 1). The final patient population consisted of 50 patients, whereof 15 patients (30%) were prospectively enrolled (part of the � Blinded for Review � cohort as described in the Methods sections), while 35 (70%) patients were retrospectively included in the study population. Twenty-two patients had a MAA, 15 were confirmed by blood culture, five by biopsy or in resected tissue; only in two cases blood culture and/or biopsy was negative, and MAA was diagnosed in an overall appraisal of clinical presentation, laboratory and imaging.
One hundred and one PET/CT scans were acquired, 50 for the initial diagnosis/baseline scan, and 51 follow up scans (22 for the first, 13 for the second, nine for the third, five for the fourth, and two for the fifth follow-up PET/CT examination). In 28 examinations (28%), an additional contrast-enhanced CT of the chest and/or the abdomen was acquired as part of the PET/CT scan (10 baseline scans, 10 first, three second, three third, and two forth PET/CT follow-up scans).
Completeness of follow-up was 92% as in four patients information could not be collected. Three patients were referred from other institutions and lost to follow-up, one patient decided to refrain from any further treatment. The remaining 46 patients (92%) were clinically followed for a median of 898 days (IQR 320-4105) after their baseline PET/CT and for a median of 429 days (IQR 82-4105) after their last PET/CT examination.
Nine (18%) of the baseline PET/CT could clearly answer the referral question, however, findings could not be confirmed, due to missing or incomplete follow-up (in these cases the impact of PET/CT on patient management was defined as "suspected"), i.e. one MAA, three rule-outs of MAA, one vascular graft infection (VGI), three lymphomas, and one endocarditis (Tables 2-5).
In nine (18%) cases, PET/CT findings were "misleading" since they were proven wrong by follow-up, i.e. in all of these cases, an MAA was suspected in PET/CT but not confirmed. The final diagnosis in these cases were: five inflammatory aneurysms, one necrotic aneurysm, one aneurysm "prerupture" (non-ruptured abdominal aortic aneurysm in a hemodynamically unstable patient), one aneurysm with a metabolically active plaque, and one aneurysm, in

PLOS ONE
which the increased metabolic activity could not be explained, but long-term clinical followup ruled out infection (Tables 2-5).
Incidental findings were detected in 12 patients, of these 11 were confirmed (five pneumonias, one lymphoma, two colorectal adenomas, one spondylodiscitis, one arteritis, one periodontitis), while one was only suspected (i.e. suspected lymphoma, lost to follow-up). In three patients, the incidental finding was the reason why PET/CT had an impact on patient management (i.e., two confirmed pneumonias and one confirmed colorectal adenoma in three patients, one with an FDG-negative aneurysm and the other two with a suspected graft infection, both with insufficient follow-up for the latter). In all other cases (n = 10), PET/CT also had an impact on management since it correctly answered the referral question (e.g. in a patient with a confirmed MAA, the additionally detected spondylodiscitis was not rated as an additional impact, but just an overall impact of PET/CT on patient management). One examination had a confirmed impact on patient management but was also misleading (i.e. staging of bronchial carcinoma was correct, however, the incidentally detected and suspected MAA was not confirmed by clinical follow-up); thus the examination was deemed "misleading" (Tables 2-5).
Hence, baseline PET/CT had impact on patient management in 82% of cases (70% confirmed, and 12% suspected). PET/CT results were misleading in 18% of cases. With regard to the referral question, a trend (P = .077) towards higher frequencies of PET/CT examinations
At follow-up, 13 (25.5%) PET/CT examinations answered the referral question conclusively (Fig 5), and all findings were confirmed by follow-up, i.e. two MAA (both progressive and confirmed by blood culture), two VGI (one progressive and one newly diagnosed, one confirmed  A and B). Both readers correctly rated the finding to be an infectious aneurysm; no gastric cancer was detected. The diagnosis was confirmed after biopsy (MAA caused by Coxiella burnetii) and the patient was treated with endovascular aortic repair and partial gastrectomy and lymphadenectomy. The patient was treated with antibiotics for 784 days and did not show any sign of recurrence at the last clinical visit 293 days after the end of antimicrobial therapy.
Another 25 (49%) of the follow-up PET/CT scans could clearly answer the referral question, however findings could not be confirmed by other means or due to missing or incomplete follow-up (in these cases the impact of PET/CT on patient management was deemed "suspected"), i.e. stable VGI (n = 7), partial regression in metabolic activity of a VGI (n = 17), complete regression but missing follow-up (n = 1).
In 13 (25.5%) cases, PET/CT findings were "misleading" as they were proven wrong by follow-up. One PET/CT showed no pathologic FDG-uptake, but the patient had endocarditis. In the remaining 12 cases a VGI was suspected in PET/CT, but not confirmed (the FDG-positive findings in these cases were attributed to inflammatory aneurysms (n = 2) or remained unclear (n = 10) (possibly postoperative or granulomatous inflammatory reaction) (Fig 6).

PLOS ONE
Five incidental findings were detected in four examinations of four patients, all of them were confirmed (one pneumonia, two colorectal adenoma, one infected intestinal fistula, and one infected pancreatic cyst). Only in one follow-up examination the incidental finding was the reason why PET/CT had an impact on patient management, i.e. one patient with a stable VGI (rated as suspected impact) with newly detected fistulas and a new colorectal adenoma (finally rated as confirmed impact) ( Tables 6-8).

Baseline and follow-up PET/CT combined
The combined evaluation of all 101 PET/CT (50 baseline PET/CT, and 51 follow-up PET/CT) demonstrated an impact on patient management in 78,5% of cases (48,5% confirmed, and 30% suspected). Results of 21,5% of the PET/CT examinations were misleading.

Discussion
The study assessed the impact of PET/CT on clinical management of patients with suspected MAA. In 101 PET/CT examinations of 50 patients, a high impact on patient management was demonstrated, which was more pronounced with baseline than with follow-up examinations. In a small subset of patients in the present study, PET/CT results were misleading. PET/CT is considered as the imaging gold standard in the diagnosis and in therapy control of many tumors. Numerous studies have demonstrated the impact of PET/CT on patient management by either questioning the referring physicians prior and after PET/CT on their intended treatment approach [1], or by comparing the additional information of PET/CT to findings of conventional imaging [11]. If, for example, a single metastasis detected only by PET/CT but not by conventional imaging, leads to an upstaging of a patient, it may cause a possible shift of strategy from local to systemic treatment. The latter is an excellent definition for "impact of PET/CT on patient management" [11]. Furthermore, if PET/CT results in oncologic imaging are equivocal or doubted, they often may be confirmed by biopsy-an option which is not viable in many infectious diseases.
To the best of the knowledge, the present study is the first study to investigate the impact of PET/CT on patient management in MAA, and only a few other studies have evaluated the impact of PET/CT in other infections [3,[12][13][14][15][16][17][18][19]. A previous study analyzed the impact of PET/CT in the management of immunocompromised patients with invasive fungal infections https://doi.org/10.1371/journal.pone.0258702.g005 [2], and defined impact by an influence in the diagnostic work-up at primary staging by assessing the extent of infection and targeting the diagnostic procedure, or by an influence on antifungal drug dosage or treatment withdrawal. In MAA, a comparable definition is difficult, since even the gold standard for the diagnosis of the disease is a combination of the overall appraisal of clinical presentation, laboratory results, and imaging [6]. Even more so, no consensus exists for determinants, or cut-offs values, which may justify the end of antimicrobial treatment in MAA. At present, treatment decisions in patients with MAA will always be based on a combination of results from different diagnostic tests and tools. Therefore, the impact of a single method (e.g., PET/CT) on patient management is almost impossible to differentiate from others. Hence, a new definition of "impact on management" for PET/CT in MAA was established. Impact was noted, when either the referral question to PET/CT was clearly answered, and clinical follow-up confirmed PET/CT results, or if and incidental finding (e.g., detection of a previously unknown site of infection or tumor) clearly changed the patient management. By these means, the present study yields considerably higher impact of PET/CT on patient management in patients with suspected MAA as compared to previous studies with oncologic patients. A recent large prospective study [1] demonstrated a change of intended management in 37% of cancer patients after PET/CT, while the present study describes a confirmed impact of at least 70% at baseline PET/CT.

Limitations
The newly introduced definition of "impact on management" challenges the comparability of the present study results to previous. However, as described above, previously used definitions for other diseases may not be easily transferred to the treatment of MAA, and no comparable previous data on the impact on management of PET/CT in MAA exist in the literature. Furthermore, no PET/CT imaging criteria for diagnosing MAA were defined, as our study was retrospective, and at the time of image reading, no criteria existed in the literature. In our previous work [7], we could show, that a cut-off criterion "higher FDG-uptake in the aneurysm than in liver background" lead to a sensitivity of 100% in diagnosing MAA, but the criterion was hampered by a low specificity due to false-positive findings (e. g. in inflammatory aneurysms).
Finally, the present study population was heterogeneous, with varying numbers of followup PET/CT examinations, using different PET/CT scanner generations, as well as prospectively and retrospectively included patients. However, it represents a large study population with MAA examined with PET/CT. Nonetheless, further studies are required to confirm the results.

Conclusion
In MAA, PET/CT has a high impact on patient management, which is more pronounced with baseline than with follow-up examinations. However, PET/CT results may be misleading in a subset of patients.
Supporting information S1