Impact of unknown incidental findings in PET/CT examinations of patients with proven or suspected vascular graft or endograft infections

Vascular graft or endograft Infections (VGEI) are rare but severe complications of vascular reconstructive surgery, and associated with significant mortality and morbidity risk. Positron emission tomography/computed tomography with 18F-fluorodeoxyglucose (PET/CT) has been shown to have a high diagnostic accuracy in the detection of VGEI. In this single-center prospective cohort study, we assessed the rate and the impact on patient management of relevant unknown incidental findings in PET/CT of patients with proven or suspected VGEI, and clinical follow-up of all patients was performed. Our study results show a comparably high rate of relevant unknown incidental findings (181 in 502 examinations), with documented direct impact on patient management in 80 of 181 (44%) of all findings. PET/CT scan- and patient-based evaluation revealed impact on patient management in 76 of 502 (17%) of all PET/CT scans, and in 59 of 162 (36%) of all patients, respectively. Furthermore, PET/CT correctly identified the final diagnosis in 20 of 36 (56%) patients without VGEI. In conclusion, in proven and suspected VGEI, PET/CT detects a high rate of relevant unknown incidental findings with high impact on patient management.

, including patients aged ≥ 18 years with any type of vascular graft operation and clinical suspicion of VGEI. The diagnosis of VGEI was defined as suggested by the Management of Aortic Graft Infection Collaboration (MAGIC). Every clinical case was adjudicated by a multidisciplinary team of infectious disease specialists, cardiovascular surgeons, radiologists, nuclear medicine physicians and microbiologists. The gold standard for VGI diagnosis was a positive microbiological culture of the deep tissue around the vascular prosthesis, obtained by open biopsy or a positive microbiological culture of an explanted vascular graft. In cases with excluded VGEI, clinical, laboratory, histopathological and imaging results were also considered, and served as the standard of reference. A case file review was performed in all patients, and information collected at the time of the initial diagnosis, at the time of imaging and at the last clinical visit (recorded until June 2020). Data included patient demographics, laboratory data, results of microbiology, clinical information, and information about treatment. Written informed consent was obtained from all participants. The institutional ethics committee approved the study, namely the Kantonale Ethikkomission Zürich (KEK-ZH-Nr. 2012-0583). All procedures were performed in accordance with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.
PET/CT data acquisition. Patients fasted for at least four hours prior to PET/CT imaging. Blood glucose levels < 12 mmol/l were accepted 22 . The standardized uptake time was 60 min in supine position. A nonenhanced CT scan for attenuation correction and anatomic localization of 18F-FDG uptake was performed with arms overhead whenever possible, using automated dose modulation (range 10-100 mA, 120-140 kV) with a scan range from the vertex of the skull to the mid of the thighs or to the feet. Two types of PET/CT scanners were used within the study period between 2012 and 2020 (i.e. Discovery VCT, and Discovery MI, both GE Healthcare, Waukesha, WI), using either the 3-dimensional acquisition mode with a fixed scan duration of two minutes per bed position (Discovery VCT), or a time-of-flight acquisition mode with a fixed scan duration of 2.5 min per bed position (Discovery MI).
PET/CT image analysis and definitions. All PET/CT were analysed independently on a dedicated workstation (Advantage Workstation, version 4.7, GE Healthcare Biosciences, Pittsburgh, PA) by two dually board-certified radiologists and nuclear medicine physicians, both with more than 8 years experience in hybrid imaging. Primary and secondary (e.g. infectious foci not in the vicinity of the aorta or other relevant findings such as malignancies) diagnoses were documented if deemed potentially relevant for patient management. A consensus reading was performed if results differed. All findings were retrospectively analysed, whether the findings were previously known (e.g.: by previous other imaging modalities). If findings were not previously known and were considered to harbour potential impact on patient management, they were defined as "unknown incidental findings".
All patients were clinically followed by reviewing electronic patient charts, to document the impact on patient management of all findings.

Statistical analyses.
Variables were expressed as median and interquartile range (25th, 75th percentiles) or percentages. All statistical analyses were performed using commercially available software (Stata/SE, Version 16.1, StataCorp, College Station, Texas).

Results
Patient population. We analyzed 505 PET/CT scans of 162 patients who consented to participate in the VASGRA Cohort Study. After dropping three PET/CT from three patients scanned for the evaluation of mycotic aneurysms prior to graft placements, we finally included 502 PET/CT of 162 patients. Patient demographics of the final study population are displayed in Table 1.

Discussion
We assessed the rate and the impact on patient management of relevant unknown incidental findings in PET/ CT of patients with proven or suspected VGEI.
Our study results show a comparably high rate of relevant unknown incidental findings (181 in 502 examinations), with documented direct impact on patient management in 80 of 181 (44%) of all findings. PET/CT scan-and patient-based evaluation revealed impact on patient management in 76 of 502 (17%) of all PET/CT scans, and in 59 of 162 (36%) of all patients, respectively. Furthermore, PET/CT correctly identified the final diagnosis in 20 of 36 (56%) patients without VGEI.
The present study represents the first study to analyze the impact of incidental findings on patient management in a large prospectively enrolled study population with suspected or proven VGEI. Several other studies have performed similar analysis in different study populations, however, the definitions of "impact on management" and "relevant unknown incidental findings" are very inhomogeneous throughout the literature. For example, Wan et al. 23 performed PET/CT in patients with psoriasis. Similar to our study, their "relevant" findings were determined by the report of the reading radiologist/nuclear medicine physician. The authors found less relevant findings (i.e. in 12% of 259 patients), and documented that the risk of discovery of significant findings was associated with age. In comparison, the age of our study population was much higher (i.e. 45 years versus 69 years) and we did not exclude patients due to other comorbidities, while Wan et al. 23 excluded patients with significant comorbidities such as diabetes or uncontrolled hypertension. Both differences may account for the higher incidence of incidental findings in our study.
On the other hand, Nihuis et al. 24 found a much higher incidence of non-melanoma findings on baseline and annual surveillance CT and PET/CT of asymptomatic melanoma patients (i.e. 912 findings in 1022 scans versus 181 findings in 502 scans in our study). The authors included findings which were not considered "relevant" findings in our study (e.g., coronary artery calcifications or heart valve calcifications), and their rate of findings with impact on patient management was lower as compared to our study (114 of 912 findings versus 80 of 181 findings). The latter, may possibly be explained by the younger age (i.e., mean age 49) of the patient population in the study by Nihuis et al. 24,25 , and also by a higher comorbidity rate in cardiovascular patients, in general 26 .
An even higher rate on impact on management was reported by Hadad et al. 27 , reporting 642 of 1090 (59%) incidental foci to be clinically relevant in a literature review of selected patient cohorts with different types of cancer. We suppose that the comparably lower rate of findings with impact on patient management in our study, Open wound with exposed graft or communicating sinus 1 , n (%) 23 (14%) Graft insertion in an infected site 1  www.nature.com/scientificreports/ may in part be due to systematic differences between patient populations with cancer and patient populations with infectious diseases. For example, an incidentally detected pneumonia in a patient with cancer will result in a change in patient management, as antibiotic treatment will be initiated. However, in a patient with a newly detected VGEI and a synchronous pneumonia, both diseases might be treated with the same antibiotic agent, rendering the incidentally detected pneumonia to "without impact on management". Hence, we believe that the rate of relevant incidental findings with impact on patient management may even be underestimated in patient populations with infectious diseases. Finally, PET/CT has been shown to have a high diagnostic accuracy in the detection of VGEI, with an excellent negative predictive value 6,7,28-30 . Our study demonstrates that PET/CT may not only rule out VGEI correctly in patients with suspected VGEI, but may also provide the correct diagnosis in a large proportion of patients without VGEI but with clinical symptoms that were responsible for the referral to PET/CT with the question for VGEI. In fact, PET/CT identified the final diagnosis in 56% of patients with clinical symptoms but without VGEI in our study. Notably, some of the final diagnoses made in the patients without VGEI and with negative PET/CT were diagnoses, which cannot reliably be detected in normal state-of-the-art PET/CT without additional intravenous contrast medium (e.g., pyelonephritis, Dressler-syndrome, or endocarditis). Our data is in line with showed diffuse FDG uptake along the ascendens graft (Index surgery: aortic arch replacement with 28 mm Intergard® prosthesis) (black arrow heads in A, white arrow heads in C) which was focally pronounced (A) indicating persistent infection. Antibiotic treatment was continued for another eight months and then successfully stopped (no signs for recurrence at the last control seven months later). The patient had known thyroiditis, which presented with diffuse FDG uptake (upper black arrow in A, white arrow in B); this finding was not evaluated for impact on patient management in our study, as it was already known prior to PET/ CT. An unknown and relevant incidental finding was detected in the sigmoid colon with intense focal FDG uptake (lower black arrow in A, white arrow in D). This PET/CT finding was rated to have impact on patient management-subsequent coloscopy and resection revealed a colonic polyp with dysplasia.  31 ] and is often superior to other imaging modalities, e.g. in fever of unknown origin 13 , chronic osteomyelitis 32 , or mycotic aortic aneurysms 15 .
Limitations of the study. The definitions of "impact on management" and "relevant unknown incidental findings" are very inhomogeneous throughout the literature and challenge the comparability of our study results to previous ones. To date, our study represents the largest prospectively enrolled study population with suspected or proven VGEI and our results clearly underline the impact of PET/CT on patient management. Notably, the study population is selective and also inhomogeneous (e.g. including aortic and peripheral grafts). Finally, it was beyond the scope of the present study to analyse the risks and drawbacks of the detection of incidental findings, as these findings may cause patient anxiety, trigger further investigations and increase health care costs 33 .

Conclusion
In proven and suspected VGEI, PET/CT detects a high rate of relevant unknown incidental findings with high impact on patient management. www.nature.com/scientificreports/

Data availability
The datasets generated during and analysed during the current study are available from the corresponding author on reasonable request.  www.nature.com/scientificreports/