Prone position PET/CT is useful in reducing gravity-dependent opacity-related [18F]fluorodeoxyglucose uptake

This study aimed to investigate whether performing [18F]fluorodeoxyglucose ([18F]FDG) positron emission tomography/computed tomography (PET/CT) in the prone position could reduce [18F]FDG uptake in dependent lungs. Patients who underwent [18F]FDG PET/CT in both supine and prone positions from October 2018 to September 2021 were reviewed retrospectively. [18F]FDG uptake of dependent and nondependent lungs was analysed visually and semi-quantitatively. A linear regression analysis was performed to examine the association between the mean standardised uptake value (SUVmean) and the Hounsfield unit (HU). A total of 135 patients (median age, 66 years [interquartile range: 58–75 years]; 80 men) were included. Dependent lungs showed significantly higher SUVmean and HU than nondependent lungs on supine position PET/CT (sPET/CT, 0.59 ± 0.14 vs. 0.36 ± 0.09, p < 0.001; − 671 ± 66 vs. − 802 ± 43, p < 0.001, respectively) and prone position PET/CT (pPET/CT, 0.45 ± 0.12 vs. 0.42 ± 0.08, p < 0.001; − 731 ± 67 vs. − 790 ± 40, p < 0.001, respectively). Linear regression analysis revealed a strong association between the SUVmean and HU in sPET/CT (R = 0.86, p < 0.001) and moderate association in pPET/CT (R = 0.65, p < 0.001). One hundred and fifteen patients (85.2%) had visually discernible [18F]FDG uptake in the posterior lung on sPET/CT, which disappeared on pPET/CT in all but one patient (0.7%, p < 0.001). [18F]FDG uptake of the lung had moderate-to-strong associations with HU. Gravity-dependent opacity-related [18F]FDG uptake can be effectively reduced on prone position PET/CT. Prone position PET/CT effectively reduces gravity-dependent opacity-related [18F]fluorodeoxyglucose uptake in the lung, potentially improving diagnostic accuracy in evaluating nodules in dependent lungs and offering a more accurate assessment of lung inflammation parameters in interstitial lung disease evaluations. • The study evaluated whether performing [18F]fluorodeoxyglucose ([18F]FDG) PET/CT could reduce [18F]FDG uptake in lungs. • In prone and supine position PET/CT, the [18F]FDG uptake and Hounsfield unit were moderately to strongly associated. • Prone position PET/CT can reduce gravity-dependent opacity-related [18F]FDG uptake by the posterior lung.


Introduction
Dependent opacity on computed tomography (CT) is an ill-defined increased subpleural attenuation that occurs in the dependent lungs and disappears when the region becomes nondependent [1].Dependent opacity can be caused by passive microatelectasis or by fluid accumulation due to gravity [2].Therefore, when imaging is performed in the supine position, gravity-dependent opacity often occurs in the posterior lung affected by gravity.
Not only gravity-dependent opacity but also high [ 18 F] fluorodeoxyglucose ([ 18 F]FDG) uptake can be seen in the dependent lungs on positron emission tomography/computed tomography (PET/CT).According to Gerbaudo et al, gravity-dependent opacity or atelectasis showed higher [ 18 F]FDG uptake than the normal lung, and there was a positive linear correlation between atelectatic lung density and [ 18 F]FDG uptake [3].Though most atelectasis showed a lower mean standardised uptake value (SUV mean ) than malignancy [3], there is an overlap between atelectasis and tumours with low [ 18 F]FDG uptake, such as small-sized tumours, lepidic-dominant adenocarcinoma, and neuroendocrine tumours [4,5].Therefore, assessments of [ 18 F] FDG uptake may be difficult when these types of lung nodules are located within the gravity-dependent opacity.In addition, gravity-dependent opacity can be a confounding factor in quantifying lung inflammation parameters such as the metabolic lung volume or total lung glycolysis on [ 18 F]FDG PET/CT [6,7].
CT scans usually obtain images in the supine position; however, CT taken in the prone position helps differentiate between gravity-dependent opacity and true lung disease [8,9].However, it is not known whether [ 18 F]FDG uptake in the dependent lung would be reduced when [ 18 F] FDG PET/CT is taken in the prone position.Therefore, we investigated whether [ 18 F]FDG PET/CT taken in the prone position could reduce [ 18 F]FDG uptake in dependent lungs.

Subjects
This study was approved by the Institutional Review Board of our institution (IRB no.2022-02-017).As this is a retrospective study, the need for obtaining informed consent was waived.In our institution, some of the patients with lung or liver lesions underwent an additional regional prone position PET/CT (pPET/CT) immediately after the routine torso supine position PET/CT (sPET/CT) because pPET/ CT is useful in reducing respiratory motion artefacts caused by diaphragmatic movement [10].The decision to perform pPET/CT was made by a nuclear medicine physician (S.H.L.) based on the previous CT or MR imaging findings before performing standard sPET/CT.We retrospectively reviewed 197 patients who underwent [ 18 F]FDG PET/CT in both supine and prone positions from October 2018 to September 2021 in our institution.Combined lung pathologies, which can affect the measurement of SUV mean or Hounsfield unit (HU) of a region of interest (ROI), were demonstrated in terms of nodules or masses, pleural effusion, interstitial lung disease, pneumonia or airway disease resulting in mosaic lung parenchymal attenuation, and consolidative lesions.Sixty-two patients were excluded due to these confounding factors (Fig. 1).

Fig. 1 Flow diagram of patient enrolment
Each patient's age, sex, smoking history, smoking duration, reason for undergoing [ 18 F]FDG PET/CT, and present illness were obtained through electronic medical records.

PET/CT imaging protocol
[ 18 F]FDG PET/CT image acquisition conditions were as follows: The patient fasted for > 6 h before [ 18 F]FDG PET/CT and was injected with [ 18 F]FDG at 5.18 MBq/kg (0.14 mCi/ kg).The blood glucose level was checked and kept below 8.33 mmol/L (150 mg/dL).After [ 18 F]FDG injection, standard sPET/CT images were acquired within 50-70 min using a Gemini TF 16 PET/CT scanner (Philips Healthcare).Initial low-dose CT (120 kVp, 50 mAs, 4 mm slice thickness) was performed first, and then PET images were obtained.PET images were reconstructed using the 3D RAMLA iterative OSEM algorithm (3 iterations, 33 subsets, no filtering), and CT-based attenuation correction was performed.Immediately after the routine torso sPET/CT, the patient changed to a prone position, and additional regional pPET/CT was performed.Image acquisition settings and reconstruction algorithms of both sPET/CT and pPET/CT were the same as above.

PET/CT image analysis
For the semi-quantitative analysis, a nuclear medicine board-certified physician (10 years of experience in PET reading and analysis) manually drew the lung ROI 2 cm above the diaphragm under the lung window setting [11] and measured the SUV mean through the workstation (Advantage Workstation 4.7, GE Healthcare).ROI of the right lung was divided into the anterior, mid, and posterior parts (Fig. 2) [12].The dependent lung refers to the posterior lung ROI in sPET/CT and the anterior lung ROI in pPET/CT, and vice versa for the nondependent lung.The HU of each ROI was also measured.Additionally, visual grading of [ 18 F]FDG uptake in the posterior lung, a gravity-dependent area in sPET/CT, was performed.For visual analysis, a 4-point scale was used with the SUV window level set from 0 to 5: grade 0, no significant [ 18 F] FDG uptake distinct from the background lung; grade 1, visually discernible [ 18 F]FDG uptake, which is higher than the background lung but lower than that of the liver [ 18 F]FDG uptake; grade 2, similar [ 18 F]FDG uptake to the liver; and grade 3, higher [ 18 F]FDG uptake than that of the liver [13,14].We also measured the tissue fraction-corrected SUV mean (SUV-TF mean ) via the previously described method suggested by Lambrou et al to exclude the air fraction of the lung because [ 18 F]FDG is not distributed in the air [7].Using this method, the SUV-TF mean can be calculated by measuring the SUV mean and HU of the lung ROI [15].
The evaluation of underlying lung parenchymal disease was performed by a chest radiologist with 15 years of experience through chest CT via the helical technique (100-120 kVp, 100-300 mAs; reconstruction was performed using a bone algorithm; 2 mm slice thickness, gantry rotation time of 0.28 s, pitch of 0.6, table speed of 23 mm per rotation, and beam width of 38.4 mm), taken on the closest date before [ 18 F]FDG PET/CT examination.Dependent atelectasis in lung parenchyma of the ROI was evaluated via low-dose chest CT taken together with [ 18 F]FDG PET/CT; however, the thin-section chest CT taken closest to [ 18 F]FDG PET/ CT was also referred.The presence of dependent atelectasis on CT images was defined as crescent-shaped ground-glass opacities representing increased vascularity or shrinkage of lung parenchymal volume or curvilinear lung parenchymal opacities perpendicular to the dependent wall of the hemithorax.The paired t-test was used to compare multiple parameters (SUVmean, HU, and SUV-TFmean) of each ROI (anterior, mid, and posterior lungs) according to their positions.The repeated ANOVA was used to compare the parameters of ROIs within the same position.The Bonferroni method was used for post hoc analysis (Bonferroni-adjusted p = 0.05/3).The linear regression analyses were used to evaluate the association between the HU and SUVmean of each ROI.[ 18 F]FDG PET/CT visual positivity according to the position was compared using the Wilcoxon signed-rank test.Except for the post hoc analysis, p value less than 0.05 was considered statistically significant.

Association between SUV mean and HU in supine and prone position PET/CT
The results of the linear regression analysis of the SUV mean according to HU on both sPET/CT and pPET/CT are shown in Fig. 3.The linear regression analysis revealed a strong association in sPET/CT between the SUV mean and HU (R = 0.86, p < 0.001).In pPET/CT, there was a moderate association between the SUV mean and HU (R = 0.65, p < 0.001) (Fig. 3).This difference in the degree of association is in line with that in Table 2.The SUV-TF mean of anterior and posterior lung ROIs did not differ significantly in sPET/CT but differed significantly in pPET/CT.

Visual [ 18 F]FDG uptake in the posterior lung according to position
Visual [ 18 F]FDG uptake in the posterior lung according to the supine and prone position is shown in Table 4. Out of 135 patients, 115 (85.2%) had visually discernible [ 18 F]FDG uptake (grades 1 or 2) on sPET/CT.However, no patient had visually discernible [ 18 F]FDG uptake on pPET/CT except for one patient (0.7%) with a persistent subpleural line.No patients showed [ 18 F]FDG uptake higher than that of the liver (grade 3) on both sPET/CT and pPET/CT.We performed subgroup analyses according to the CT findings.Of the 135 patients, 77 (57.0%) had dependent opacity or atelectasis on chest CT in the supine position.Among them, nine patients (11.7%) showed high [ 18 F]FDG uptake (grade 2, similar to the liver) on sPET/CT, which is high enough to interfere with the evaluation of small lung nodules (Fig. 4).In eight of the nine patients, [ 18 F]FDG uptake disappeared on pPET/CT, and [ 18 F]FDG uptake decreased from grade 2 to 1 in one patient with a persistent subpleural line.The remaining 58 patients (43.0%) had no dependent opacity or atelectasis on chest CT and did not have high [ 18 F]FDG uptake (grade 2 or 3).

Discussion
In the atelectatic lung, the SUV mean may increase due to the dense alveolar structure of the collapsed lung [16].Because the alveolar size of the posterior lung is smaller than that of the anterior lung in the supine position [17], the attenuation of the dependent lung is higher than that of the nondependent lung [11].As demonstrated by the linear regression analysis, [ 18 F]FDG uptake is proportional to HU and is higher in the dependent lung than in the nondependent lung.When the air fraction was removed, there was no significant difference in the SUV mean (SUV-TF mean ) between the dependent and nondependent lungs in sPET/CT.Therefore, we think gravity-dependent opacity-related [ 18 F]FDG uptake is a secondary finding due to increased attenuation rather than other factors such as inflammation.[ 18 F]FDG uptake in any lung ROI was not higher than that of the liver, and the SUV-TF mean was ~ 2.0, which is similar to that of other solid organs, which supports our hypothesis.
Since [ 18 F]FDG uptake is proportional to HU, it is reasonable to think that gravity-dependent opacity-related [ 18 F] FDG uptake would decrease on pPET/CT.As we expected, [ 18 F]FDG uptake in the posterior lung was significantly decreased on pPET/CT in both semi-quantitative and visual analyses.In particular, all high [ 18 F]FDG uptake similar to that of the liver in nine patients on sPET/CT decreased to lower than that of the liver on pPET/CT.
Although SUV mean and HU were higher in the dependent lung than in the nondependent lung on both sPET/CT and pPET/CT, the difference was lesser in pPET/CT (Table 3).Unlike the smaller alveolar size of the dependent lung in the supine position, the alveolar size becomes more uniform in the prone position [17].Therefore, the SUV mean and HU on pPET/CT also seem to be more uniform than they are on sPET/CT.
On pPET/CT, the association between SUV mean and HU was weaker than that on sPET/CT, and the SUV-TF mean of the dependent and nondependent lungs differed significantly.This could be due to different proportions of lung tissues and blood between the two positions.The density of the lung is determined by the relative proportions of air, blood, and lung tissues [18].Among these, Lambrou's method used in this study removes only the air fraction [7].Because the SUV mean of blood and lung tissues may differ, the SUV-TF mean may be affected by the proportion of blood flowing to lung tissues.In general, solid organs such as the liver, kidney, spleen, and pancreas have a higher SUV mean than the mediastinal blood pool [13,19].Therefore, SUV-TF mean might be high when the lung tissue proportion is high, and it might be low when the blood proportion is high.Even if the SUV-TF mean of the dependent and nondependent lungs are similar on sPET/CT due to similar proportions of lung tissues and blood, the SUV-TF mean of the dependent lungs of pPET/CT may be lowered because the gravity-induced increase in blood perfusion may increase the blood proportion.Further studies are needed to prove this hypothesis.
Our study has two important clinical implications.First, when evaluating nodules located in the posterior lung, [ 18 F]FDG uptake of nodules may be indistinguishable from [ 18 F]FDG uptake caused by dependent atelectasis.In our study, 11.7% of patients with gravity-dependent opacity shown on CT had [ 18 F]FDG uptake similar to that of the liver.Small nodules or nodules with low [ 18 F] FDG uptake, such as lepidic-dominant adenocarcinoma, neuroendocrine tumours [4,5], metastatic nodules from hepatocellular carcinoma, or renal cell carcinoma [20,21], may not have [ 18 F]FDG uptake that is distinct from the background lung activity.While performing pPET/ CT, gravity-dependent [ 18 F]FDG uptake can be reduced and respiratory motion artefacts of the posterior lung can be avoided [10] to properly evaluate the posterior lung nodule, as shown in Fig. 4. Second, when measuring parameters regarding lung inflammation in [ 18 F] FDG PET/CT, pPET/CT may be more accurate.Several studies have reported that lung inflammation parameters measured in [ 18 F]FDG PET/CT of interstitial lung disease reflect the disease activity and prognosis [22][23][24][25][26].However, it is well known that gravity-dependent opacity can mimic early ILD, and prone position chest CT is sometimes performed if necessary [8].Therefore, when measuring lung inflammation parameters in sPET/CT, gravity-dependent [ 18 F]FDG uptake can be a confounding factor because there is an anteroposterior gradient in the lung.pPET/CT may be more appropriate because [ 18 F] FDG uptake is more uniform.
This study has several limitations.First, respiratory gating was not available in our institution.In PET/CT, respiratory misregistration of CT and PET may occur, especially in sPET/CT [10], so tissue fraction correction using SUV mean and HU may be inaccurate.Second, the lung ROIs of sPET/CT and pPET/CT may not exactly match each other because the shape of the lungs changes with the patient's position [27].
In conclusion, lung [ 18 F]FDG uptake and HU have a moderate-to-strong association on PET/CT, and pPET/CT can effectively reduce gravity-dependent opacity-related [ 18 F]FDG uptake.

Fig. 2
Fig. 2 Divisions of the lung on [ 18 F]fluorodeoxyglucose ([ 18 F]FDG) PET/CT.The posterior lung (P) was posterior to the anterior margin of the vertebral body.The remaining lung was divided in half based on the anteroposterior axis and marked as anterior (A) and mid (M)

Fig. 3
Fig.3Linear regression analysis between standardised uptake value (SUV mean ) and the Hounsfield unit (HU).A higher correlation between HU and SUV mean was observed on supine position PET/CT (R = 0.86, p < 0.001) than on prone position PET/CT (R = 0.65, p < 0.001)

Fig. 4
Fig. 4 Gravity-dependent opacity-related [ 18 F]fluorodeoxyglucose ([ 18 F]FDG) uptake surrounding a lung nodule.A 60-year-old man with a 14-mm pulmonary nodule in the right lower lobe underwent [ 18 F]FDG PET/CT.The [ 18 F]FDG uptake of the lung nodule could not be clearly identified from the background lung due to gravitydependent opacity-related [ 18 F]FDG uptake similar to the liver on Statistical analyses were performed using IBM SPSS Statistics for Windows (version 27; IBM Corp.), STATA version 17.0 (StataCorp, College Station), and MedCalc Statistical Software version 19.2.1 (MedCalc Software Ltd; https:// www.medca lc.org;2020).

Table 2
Characteristics of chest CT and [ 18F]fluorodeoxyglucose PET/CT findings according to the position Data are expressed as the mean ± standard deviation SUV mean , mean standardised uptake value; HU, Hounsfield unit; SUV-TF mean , tissue fraction-corrected SUV mean * p < 0.05 was considered statistically significant