Toward Reliable Uptake Metrics in Large Vessel Vasculitis Studies

The aim of this study is to investigate the influence of sex, age, fat mass, fasting blood glucose level (FBGL), and estimated glomerular filtration rate (eGFR) on blood pool activity in patients with large vessel vasculitis (LVV). Blood pool activity was measured in the superior caval vein using mean, maximum, and peak standardized uptake values corrected for body weight (SUVs) and lean body mass (SULs) in 41 fluorodeoxyglucose positron emission tomography/computed tomography (FDG-PET/CT) scans of LVV patients. Sex influence on the blood pool activity was assessed with t-tests, while linear correlation analyses were used for age, fat mass, FBGL, and eGFR. Significantly higher SUVs were found in women compared with men, whereas SULs were similar between sexes. In addition, higher fat mass was associated with increased SUVs (r = 0.56 to 0.65; all p < 0.001) in the blood pool, but no correlations were found between SULs and fat mass (r = −0.25 to −0.15; all p > 0.05). Lower eGFR was associated with a higher FDG blood pool activity for all uptake values. In FDG-PET/CT studies with LVV patients, we recommend using SUL over SUV, while caution is advised in interpreting SUV and SUL measures when patients have impaired kidney function.


Introduction
Semi-quantitative measurements, e.g., the mean or maximum standardized uptake value (SUV), are increasingly being used and recommended in 2-[fluorine-18]-fluoro-2deoxy-D-glucose positron emission tomography (FDG-PET). The SUV is a simple, unitless metric in which the amount of activity within a volume of interest (VOI) in the PET scan is corrected for the injected radiotracer dose and for body weight. In 1993, Zasadny et al. recommended to use the SUV normalized to lean body mass (LBM), referred to as SUL, due to the SUV's dependency on body weight [1]. FDG uptake in fat is low, whereas body weight highly depends on fat [2]. This would imply that correcting for body weight is inadequate and might lead to different treatment approaches between patients with different fat percentages or to erroneous patient monitoring when fat mass changes. Thus, where Act VOI is the activity concentration measured in the VOI; the Act administered is the net administered activity corrected for physical decay of FDG to the start of the acquisition; and 1 mL of tissue is assumed to weigh 1 g [3].
The most recent EANM guideline for oncological imaging recommends quantifying the tracer uptake by using SUL [3]. However, in PET imaging of inflammatory disorders, such as large vessel vasculitis (LVV), there is no consensus on the best correction method of FDG uptake [4,5]. Most (recent) LVV PET studies only used SUV as quantitative scoring method and did not mention SUL [6][7][8][9].
Furthermore, recent literature demonstrated differences in glucose metabolism with sex and age, possibly influencing the uptake activity during scanning [10]. Besides, impaired renal function slows down the clearance of FDG and thus may increase the blood pool activity at the standard 60 min acquisition time [11]. Guidelines consider this issue when intravenous contrast material has to be applied, but do not advise for FDG [3,4].
Although these semi-quantitative measurements are mainly used in research settings of patients with LVV so far, it is of utmost importance that these values are correct and reliable to potentially use FDG PET to achieve optimal diagnosis, therapy monitoring, and consistent treatment strategies. Therefore, the aim of this study was to investigate the influence of sex, age, fat mass, fasting blood glucose level (FBGL), and kidney function on blood pool SUVs and SULs in patients with large vessel vasculitis.

Patients and Scan Acquisition
In this study, electronic patient files of patients with LVV were retrospectively checked for the presence of a FDG PET/CT scan before start of treatment. All patients signed informed consent as part of a prospective cohort study, which was approved by the institutional review board of the UMCG (METc 2010/222) [12]. The LVV diagnosis was made by a rheumatologist. This diagnosis had to be manifested for at least 6 months after initial presentation.
All scans were performed using an integrated PET/CT system (Biograph mCT 40 or 64-slice or Vision; Siemens, Knoxville, TN, USA). The FDG-PET/CT procedure was performed according to the EANM guidelines, which included at least 6 h fasting prior to FDG injection (3 MBq/kg), subsequent 60 min waiting, and image acquisition of 3 min per bed position. All scans were reconstructed according to EARL for semi-quantitative analysis [13].

Quantification
Blood pool activity was measured by drawing VOIs within the boundaries of the superior caval vein (SCV) using Hermes Affinity Viewer v2.02 software (Hermes Medical Solutions Inc., Greenville, NC, USA). The SCV was manually delineated on all low-dose CT slices where the SCV was visible. Manual delineation started cranially at the lower border of the first right costal cartilage (where the left and right brachiocephalic veins end). The SCV was delineated until the first CT slice where the SCV was not distinguishable anymore from the right atrium. After overlaying the co-registered PET image with the low-dose CT, voxels with FDG-uptake spillover from neighboring tissue (i.e., heart tissue or the aortic wall) were carefully excluded from the SCV VOI (see Figure 1 for an example). Delineation was performed by an MD/PhD student with three years of experience with vessel segmentation in PET/CT. The SCV was segmented a second time in a random sample of 10 patients (25%) in order to assess intra-observer reliability. distinguishable anymore from the right atrium. After overlaying the co-registered PET image with the low-dose CT, voxels with FDG-uptake spillover from neighboring tissue (i.e., heart tissue or the aortic wall) were carefully excluded from the SCV VOI (see Figure  1 for an example). Delineation was performed by an MD/PhD student with three years of experience with vessel segmentation in PET/CT. The SCV was segmented a second time in a random sample of 10 patients (25%) in order to assess intra-observer reliability. , and the PET/CT with the SCV manually delineated, while carefully excluded the spillover from the heart next to it (c).
The mean, peak, and maximum standardized uptake value (SUVmean and SULmean, SUVpeak and SULpeak, and SUVmax and SULmax) of the SCV were chosen as activity parameters. LBM was calculated according to Janmahasatian et al. [14] for male and female, respectively, as recommended in the EANM guidelines for oncological imaging [

Statistical Analysis
Correlation analyses were performed comparing the activity parameters in the SCV according to age, FBGL (mmol/L), fat mass (= body weight − LBM), and estimated glomerular filtration rate (eGFR) calculated using the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) equation [15]. Pearson's r was calculated when activity parameters and possible influencing factors were normally distributed. Spearman's r was calculated when activity parameters and possible influencing factors were not normally distributed. Student's t-test or Mann-Whitney U test was done to compare groups (i.e., male vs. female) after testing for normality. After Holm-Bonferroni correction, results were considered statistically significant when p < 0.05. An intraclass correlation coefficient (two-way random model) was calculated to assess the intra-observer reliability. Coefficients between 0.75 and 1.00 were considered excellent, between 0.60 and 0.74 good, between 0.40 and 0.59 fair, and below 0.  The mean, peak, and maximum standardized uptake value (SUV mean and SUL mean , SUV peak and SUL peak , and SUV max and SUL max ) of the SCV were chosen as activity parameters. LBM was calculated according to Janmahasatian et al. [14] for male and female, respectively, as recommended in the EANM guidelines for oncological imaging [3]:

Statistical Analysis
Correlation analyses were performed comparing the activity parameters in the SCV according to age, FBGL (mmol/L), fat mass (= body weight − LBM), and estimated glomerular filtration rate (eGFR) calculated using the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) equation [15]. Pearson's r was calculated when activity parameters and possible influencing factors were normally distributed. Spearman's r was calculated when activity parameters and possible influencing factors were not normally distributed. Student's t-test or Mann-Whitney U test was done to compare groups (i.e., male vs. female) after testing for normality. After Holm-Bonferroni correction, results were considered statistically significant when p < 0.05. An intraclass correlation coefficient (two-way random model) was calculated to assess the intra-observer reliability. Coefficients between 0.75 and 1.00 were considered excellent, between 0.60 and 0.74 good, between 0.

Results
Patients were selected from a prospective cohort with newly diagnosed LVV patients. In total, 43 patients from this cohort underwent an FDG-PET/CT scan at time of diagnosis before start of treatment between 2011 and 2020, and were included in this study. Two patients were excluded because of missing eGFR and height data. Patient characteristics are presented in Table 1. Patient characteristics were equal between female and male patients, except for LBM and fat mass, which were, respectively, lower and higher in females than males. First, we investigated whether the patient demographics of sex and age influenced the blood pool activity ( Figure 2). Blood pool activity measured as SUV mean , SUV peak , and SUV max was significantly higher in females compared with males (p = 0.016, p = 0.010, and p = 0.010, respectively). However, when measured as SUL mean , SUL peak , and SUL max , no significant differences were found (p = 0.087, p = 0.195, and p = 0.114, respectively). No association between age and blood pool activity was found (Table 2 and Figure 3). Table 2. Pearson's correlation coefficients for age, fat mass, FBGL, and eGFR, with the PET activity parameters. The 95% confidence intervals are shown between brackets. Significance of the correlations is denoted by * (p < 0.05), ** (p < 0.01), *** (p < 0.001), and **** p < 0.0001).

Age
Fat Mass FBGL eGFR Increased fat mass, determined by subtracting the LBM from the total body mass, was associated with increased SUV mean , SUV peak , and SUV max in the blood pool ( Figure 4). When correcting for lean body mass by using SUL, no significant correlations were found with fat mass ( Table 2) FBGL was not associated with FDG activity in the blood pool (Table 2 and Figure 5). Conversely, decreased kidney function (in eGFR) was associated with a higher FDG activity in the blood pool (Table 2 and Figure 6). Numerical presentation of the uptake metrics shown in the graphs may be found in the Appendix A.
Intraclass correlation coefficients of two separate measurements by the same observer yielded excellent or good coefficients of 0.86 for SUV mean , 0.71 for SUV peak , 0.81 for SUV max , 0.88 for SUL mean , 0.69 for SUL peak , and 0.83 for SUL max . Sex differences in blood pool activity as measured by the mean standardized uptake value (SUV mean ), peak standardized uptake value (SUV peak ), maximum standardized uptake value (SUV max ), mean standardized uptake value corrected for lean body mass (SUL mean ), peak standardized uptake value corrected for lean body mass (SUL peak ), and maximum standardized uptake value corrected for lean body mass (SUL max ). Results from t-tests are shown in the graphs, with * = p < 0.05 and ns = not significant.  Increased fat mass, determined by subtracting the LBM from the total body mass, was associated with increased SUVmean, SUVpeak, and SUVmax in the blood pool ( Figure 4). When correcting for lean body mass by using SUL, no significant correlations were found with fat mass (Table 2)   . Plots of the patients' ages in years and blood pool activity as measured by the mean standardized uptake value (SUV mean ), peak standardized uptake value (SUV peak ), maximum standardized uptake value (SUV max ), mean standardized uptake value corrected for lean body mass (SUL mean ), peak standardized uptake value corrected for lean body mass (SUL peak ), and maximum standardized uptake value corrected for lean body mass (SUL max ). Pearson's r values are shown in the upper left corners. None of the correlations were statistically significant (p = 0.582, p = 0.785, p = 1.000, p = 0.323, p = 1.000, p = 0.550 for SUV mean , SUV peak , SUV max , SUL mean , SUL peak , SUL max , respectively). Diagnostics 2021, 11, x FOR PEER REVIEW 8 of 14 Figure 4. Plots of fat mass in kilograms and blood pool activity as measured by the mean standardized uptake value (SUVmean), peak standardized uptake value (SUVpeak), maximum standardized uptake value (SUVmax), mean standardized uptake value corrected for lean body mass (SULmean), peak standardized uptake value corrected for lean body mass (SULpeak), and maximum standardized uptake value corrected for lean body mass (SULmax). FBGL was not associated with FDG activity in the blood pool (Table 2 and Figure 5). Conversely, decreased kidney function (in eGFR) was associated with a higher FDG activity in the blood pool (Table 2 and Figure 6). Numerical presentation of the uptake metrics shown in the graphs may be found in the Appendix A.
Intraclass correlation coefficients of two separate measurements by the same observer yielded excellent or good coefficients of 0.86 for SUVmean, 0.71 for SUVpeak, 0.81 for SUVmax, 0.88 for SULmean, 0.69 for SULpeak, and 0.83 for SULmax.  . Plots of fat mass in kilograms and blood pool activity as measured by the mean standardized uptake value (SUV mean ), peak standardized uptake value (SUV peak ), maximum standardized uptake value (SUV max ), mean standardized uptake value corrected for lean body mass (SUL mean ), peak standardized uptake value corrected for lean body mass (SUL peak ), and maximum standardized uptake value corrected for lean body mass (SUL max ). Fat mass was calculated by subtracting the lean body mass from the total body weight. Pearson's r values are shown in the upper left corners, with p < 0.001 (SUV mean ), p < 0.0001 (SUV peak ), p < 0.0001 (SUV max ), and no statistical significance for SUL mean (p = 0.346), SUL peak (p = 0.362), and SUL max (p = 0.366).   . Plots of fasting blood glucose level (FBGL) in mmol/L and blood pool activity as measured by the mean standardized uptake value (SUV mean ), peak standardized uptake value (SUV peak ), maximum standardized uptake value (SUV max ), mean standardized uptake value corrected for lean body mass (SUL mean ), peak standardized uptake value corrected for lean body mass (SUL peak ), and maximum standardized uptake value corrected for lean body mass (SUL max ). Pearson's r values are shown in the upper left corners. None of the correlations were statistically significant (p = 1.000, p = 1.000, p = 0.735, p = 1.000, p = 0.974, p = 1.000 for SUV mean , SUV peak , SUV max , SUL mean , SUL peak , SUL max , respectively). Figure 6. Plots of the estimated glomerular filtration rate (eGFR) and blood pool activity as measured by the mean standardized uptake value (SUVmean), peak standardized uptake value (SUVpeak), maximum standardized uptake value (SUVmax), mean standardized uptake value corrected for lean body mass (SULmean), peak standardized uptake value corrected for lean body mass (SULpeak), and maximum standardized uptake value corrected for lean body mass (SULmax).

Discussion
Guidelines for [18F]FDG-PET/CT imaging in oncological diseases recommend the use of SUL for semi-quantitative analysis instead of the commonly used SUV [3]. In inflammatory diseases there is no consensus regarding which parameter to use for analysis. Therefore, we investigated different quantitative measurements (SUV and SUL) SULpeak r = -0.3482 Figure 6. Plots of the estimated glomerular filtration rate (eGFR) and blood pool activity as measured by the mean standardized uptake value (SUV mean ), peak standardized uptake value (SUV peak ), maximum standardized uptake value (SUV max ), mean standardized uptake value corrected for lean body mass (SUL mean ), peak standardized uptake value corrected for lean body mass (SUL peak ), and maximum standardized uptake value corrected for lean body mass (SUL max ). Pearson's r values are shown in the upper left corners, with p = 0.009 (SUV mean ), p = 0.035 (SUL mean ), p = 0.024 (SUV peak ), p = 0.042 (SUL peak ), p = 0.042 (SUV max ), p = 0.026 (SUL max ).

Discussion
Guidelines for [18F]FDG-PET/CT imaging in oncological diseases recommend the use of SUL for semi-quantitative analysis instead of the commonly used SUV [3]. In inflammatory diseases there is no consensus regarding which parameter to use for analysis. Therefore, we investigated different quantitative measurements (SUV and SUL) for defining the blood pool activity in a specific patient group with LVV. Besides a higher fat mass in female patients compared with male patients, the results showed a significantly higher SUV mean , SUV max , and SUV peak in women compared with men. When corrected for LBM, no significant differences were found (for SUL mean , SUL max , and SUL peak ). Furthermore, a significant positive correlation between SUVs and fat mass was found, meaning that a higher fat mass may slow down or reduce the FDG uptake in organs and tissues, thereby subsequently increasing the blood pool FDG activity. To better understand the underlying mechanisms, exact interactions should be investigated in future research. When corrected for LBM (SUL), no correlation was noted. Additionally, no association was found between FBGL and blood pool activity, whereas a negatively proportional correlation was found between eGFR and blood pool activity, meaning that impaired kidney function may result in higher blood pool activity values through reduced renal FDG clearance.
To compare between patients and for consistent treatment strategies it is of utmost importance that imaging metrics correctly reflect the severity of the disease. The population variation of LVV patients, including GCA and Takayasu's arteritis, is large [16]. Large differences in body weight and amount of body fat between and within patients over time may exist, implying that SUV measurements might lead to erroneous interpretation of the (follow-up) results. Our results, although with a small sample size, strongly suggest that the amount of body fat influences the blood pool activity, leading to differences between females and males in SUVs. This could be explained by the fact that, in general, females have a higher percentage of body fat relative to their body weight than men [17]. SUL could thus be a more stable metric, also supported by our findings. This is in line with previous research from Sarikaya et al. which found that SUV overestimates metabolic activity in all patients, but even more in obese patients [18]. SUL, however, is not affected by body fat, and thus more accurate.
Besides SUL, the target-to-background ratio (TBR) is also a metric that may be more reliable in some diseases, especially in therapy follow up studies, and is recommended by guidelines for LVV imaging using blood pool as background [4]. Because TBR is calculated by dividing the SUV or SUL in the target tissue by the SUV or SUL in the background, the correction methods are also divided by each other and thus do not play a role anymore: TBR = SUV or SUL(VOI) SUV or SUL(background) .
Contradictory to previous research [19][20][21], we did not find any association between FBGL and blood pool activity. This could be explained by the low variance in FBGL of our patient dataset; over 80% of the patients had an FBGL level lower than 7.0 mmol/L, which was proposed as upper cut-off value whether to perform the [18F]FDG-PET scan at that moment by Bucerius et al. [19].
We did find a negative correlation between blood pool activity values and eGFR. This is in line with results from previous studies. Rosenblum et al. showed in both Takayasu's and GCA patients a negative correlation between blood pool activity and GFR during 1-h imaging [22]. Derlin et al. demonstrated in 50 [18F]FDG-PET scans a significant negative correlation between blood pool SUVs and eGFR [23]. Both studies warn that this could result in overcorrection when TBRs are calculated, as GFR was not associated with FDG uptake in the arterial wall [22,23]. In the same study, Rosenblum et al. showed that at 2-h imaging blood pool activity and TBRs were not associated with GFR anymore [22]. Furthermore, Toriihara et al. found significantly higher uptake values in soft tissues, spleen, and blood pool in chronic renal failure patients compared with the control group with normal kidney function [24]. Laffon et al. even proposed to use a lower [18F]FDG radiation dose in patients with renal failure by using their two-compartment model, which would result in similar uptake values due to the decreased clearance [25]. As stated before, current guidelines do not address this topic concerning the use of [18F]FDG. Some of the abovementioned problems may be eliminated by dynamic or dual time point PET, which should be investigated for LVV patients in future research [26,27]. However, as this is currently not clinical practice yet, based on our results, we suggest being cautious with interpreting uptake measurements when patients have impaired renal function.
Our study has some limitations. First, the number of patients included was relatively small. To make recommendations for specific cutoff values in renal function or lower administration of radiotracer dose for example, a larger sample size would be necessary. However, our results coincide with previous research in other diseases and therefore show the importance of consistent scoring in LVV. Second, the variance in FBGL was small and levels itself were mostly low compared with previous proposed cutoff values. Last, only one author delineated the SCVs and therefore interobserver variability analysis was not done. However, intra-observer reliability was rated excellent or good.
In conclusion, fat mass has a significant influence on blood pool SUVs, resulting in differences between sexes. SUL scores are not dependent of the amount of body fat in a patient and thus result in more realistic scores and potentially more consistent treatment strategies than SUV measurements. In addition, eGFR influences blood pool activity in a negative proportional manner. Therefore, we recommend using SUL instead of SUV as quantitative measurement in [18F]FDG-PET scans of patients with LVV and to be cautious with the interpretation of quantitative measurements when patients have impaired renal function.  Informed Consent statement: Informed consent was obtained from all subjects involved in the study.

Data Availability Statement:
The data are not publicly available due to the still ongoing prospective study.