Quantification and Kinetic Analysis of Brain PET Studies

Chapter 7 Summary, general discussion and future perspectives Summary Summary, general discussion and future perspectives 7.1 Summary In this thesis various quantitative aspects of positron emission tomography (PET) studies were assessed. The first part investigated some PET instrumentation related aspects, focussing mainly on the performance of a high resolution brain PET scanner, the HighResolution Research Tomograph (HRRT). This performance evaluation was followed by an assessment of the impact of analysis methods, reconstruction settings and repositioning on the quantification of oncology and brain PET studies. Finally, an optimal pharmacokinetic model was developed for analysis of [11C]phenytoin studies in humans. More specifically, in Chapter 2, a direct comparison between HRRT (high-resolution brain) and HR+ (standard whole-body) PET only scanners for quantitative brain studies using three tracers with vastly different tracer distributions was performed. After resolution matching, HRRT-derived kinetic parameter values correlated well with those of HR+ for all tracers (intra-class correlation coefficients ≥0.78), having a good absolute inter-scanner test-retest variability (≤15%). However, systematic differences can be seen for HRRT-derived kinetic parameter values (range =13 to +15%). Quantification of kinetic parameters based on plasma-input models leads to comparable results when spatial resolution between HRRT and HR+ data is matched. When using reference-tissue models, differences remain that are likely caused by differences in attenuation and scatter corrections and/or image reconstruction. In Chapter 3, the impact of different scatter correction strategies on quantification of HRRT data for three tracers covering a wide range in kinetic profiles were evaluated. To reduce the effects of patient motion on scatter scaling factors, a margin in the attenuation correction factor (ACF) sinogram was applied prior to 2D or 3D single scatter simulation (SSS). Some (R)-[11C]verapamil studies showed prominent artefacts that disappeared with an ACF-margin of 10 mm or more. Use of 3D SSS for (R)-[11C]verapamil showed a statistically significant increase in volume of distribution compared with 2D SSS (p<0.05), but not for [ 11C]raclopride and [11C]flumazenil studies (p>0.05). When there is a patient motion-induced mismatch between transmission and emission scans, applying an ACF-margin resulted in more reliable scatter scaling factors but did not change (and/or deteriorate) quantification. Chapter 4 describes the assessment of PET/CT repeatability dependencies on reconstruction settings, analysis methods, scan duration (or image noise) and phantom position in the field of view (FOV). In this study, multiple (repeated) scans have been performed using a 68Ge uniformity phantom, a NEMA image quality (IQ) phantom and a 3D Hoffman brain phantom filled with 18F solution. Studies were performed with and without repositioning the phantom and all scans (12 replicates for IQ phantom and 10 replicates for Hoffman brain phantom) were performed using equal count statistics. For the 68Ge uniformity phantom, the coefficient of variation (COV%) and variance of the voxel values across the phantom were studied as function of scan statistics and reconstruction settings. For the NEMA IQ phan110 Summary, general discussion and future perspectives Summary tom we studied the maximum, peak and mean recovery coefficients (RC) in each sphere as function of experimental conditions (noise level, reconstruction settings and phantom repositioning). For the 3D Hoffman phantom the mean activity concentration was determined within several volumes of interest and activity recovery and its precision was studied as function of experimental conditions. As a result, we found that for all phantom studies voxel noise (expressed by variance and COV) and repeatability of RC (RCmax or RCmean) depended on reconstruction settings and frame duration, as expected. When exploring RCmax, RCpeak or RCmean of the spheres in the NEMA IQ phantom, it was observed that repeatability depended largely on phantom position, with RCmax being most (up to 30%, p<0.05) and RCpeak being least (up to 9%, p>0.05) sensitive to phantom repositioning. Moreover, impact of phantom repositioning on SUV metrics also depended on sphere size. For the brain phantom, regional average activity concentrations or RC were only minimally affected by phantom repositioning (<5% repeatability). The repeatability of quantitative PET metrics depends on the combination of reconstruction settings, data analysis methods and scan duration (scan statistics). Moreover, repeatability was also affected by phantom repositioning but its impact depended largely on the data analysis method (max, peak or mean) being used. In conclusion, the study suggests that for oncological PET studies use of SUVpeak may be preferred over SUVmax because SUVpeak seems less sensitive to patient positioning/tumor sampling effects. Chapter 5 describes the assessment of an optimal plasma kinetic model for quantification of a novel P-glycoprotein (Pgp) tracer, [11C]phenytoin. Overexpression of Pgp is thought to be an important mechanism of pharmacoresistance in epilepsy. Recently, [11C]phenytoin has been evaluated preclinically as a tracer for Pgp. The aim of the study was to assess the optimal plasma kinetic model for quantification of [11C]phenytoin studies in humans using non-linear regression (NLR) analysis of time activity curves. Dynamic [11C]phenytoin PET scans of 6 healthy volunteers with arterial sampling were acquired twice on the same day and analyzed using single- and 2-tissue-compartment models with and without a blood volume parameter. Global and regional test–retest (TRT) variability was determined for both plasma to tissue rate constant (K1) and volume of distribution (VT). According to the Akaike information criterion, the reversible single-tissue-compartment model with blood volume parameter was the preferred plasma input model. Mean TRT variability ranged from 1.5% to 16.9% for K1 and from 0.5% to 5.8% for VT. Larger VOI showed better repeatabilities than smaller regions. A 45 min scan provided essentially the same K1 and VT values as a 60 min scan. A reversible single-tissue-compartment model with blood volume seems to be a good candidate model for quantification of dynamic [11C]phenytoin studies. Scan duration may be reduced to 45 min without notable loss of accuracy and precision of both K1 and VT, although this still needs to be confirmed under pathologic conditions. In Chapter 6, the performance of various methods for generating quantitative parametric 111 General discussions and future perspectives Summary, general discussion and future perspectives images of dynamic [11C]phenytoin PET studies were evaluated. Parametric images were generated using Logan plot analysis, a basis function method (BFM) and spectral analysis (SA). Parametric VT and K1 were compared to those obtained from NLR. In addition, global and regional TRT variability was determined for parametric K1 and VT values. Biases in VT observed with all parametric methods were less than 5%, For K1, SA showed negative bias of 16%. Mean TRT variabilities of VT and K1 were less than 10% for all methods. Shortening the scan duration to 45 min provided similar VT and K1 with comparable TRT performance compared to 60 min data. Among the various parametric methods tested, BFM provided parametric VT and K1 values with the least bias compared to NLR data and showed TRT variabilities lower than 5%, also for smaller VOI sizes (i.e. higher noise levels) and shorter scan duration. 7.2 General discussions and future perspectives The first part of this thesis compared the performance of HRRT scanner to the HR+ scanner using three different tracers with vastly different tracer distributions. The HRRT has a higher intrinsic spatial resolution compare to HR+ which results in smaller partial-volume errors than seen in the clinical PET system (HR+). Even though quantification of the HRRT scanner is comparable to HR+, there is still a room for improvement in reconstruction (settings), and attenuation and scatter correction, especially for quantification based on reference tissue models. The HRRT is no longer commercially available. Nevertheless, the findings and recommended improvements to the software may also be applicable to the more modern clinical multimodality PET scanners, e.g. PET/CT and PET/MRI. As research is aimed at improving the spatial resolution of PET scanners, the acquired data will likely be sparse and, consequently, may suffer from bias using iterative reconstructions with nonnegativity constraints. In addition, since bone tissue cannot easily be accounted for during MRI-based attenuation correction, PET/MRI quantification may be hampered in the vicinity of the skull. Therefore, validating modern PET systems against more quantitatively robust PET systems using clinical studies with various tracer distributions is of utmost importance, particularly for quantitative studies. When there is a mismatch between transmission and emission scans due to patient motion, applying an ACF-margin results in more reliable scatter scaling factors but does not change (and/or deteriorate) quantification. Current and future state-of-the-art clinical PET/CT systems may be equally sensitive to inaccurate scatter scaling factors (e.g. due to patient motion) when an SSS algorithm is used, which is similar to that currently in use for the HRRT. Therefore, a relatively easy to implement, fast and low cost method, such as the one presented here, to avoid scatter scaling issues when low levels of patient motion are present is relevant for all PET systems. 112 Summary, general discussion and future perspectives General discussions and future perspectives As for the HRRT, there are various alternatives to the current tail-fitting SSS algorithm for estimating scatter. An accurate way to estimate scatter would be to perform a full Monte Carlo (MC) simulation. However, such a method may be computationally expensive. Recently, a new hybrid scatter correction strategy has been proposed for PET/CT studies to overcome scatter scaling issues due to incorrect tail fitting. This method uses SSS to approximate the shape of the scatter contribution but scales it based on a low-count MC simulation. Although initial results are promising, MC-SSS is sensitive to outside FOV scatter as well, since it estimates the scatter scaling factor per bed position (i.e. not plane by plane). Another promising new alternative avoiding spatial mismatch between emission and transmission data and thereby scatter scaling issue is the use of the so-called maximum likelihood for activity and attenuation (MLAA) reconstruction algorithm. MLAA is a promising reconstruction algorithm that could enable future time-of-flight (TOF) PET/CT and PET/MR scanners to estimate attenuation and scatter without the need for a transmission scan. Furthermore, MLRR [33], a recent reconstruction algorithm that iteratively reconstructs the activity image while registering the available attenuation image, so that the pair of activity and attenuation images maximize the likelihood of the TOF emission sinogram, could also compensate for a patient-induced mismatch between transmission and emission data. Unfortunately, the HRRT is not capable of TOF measurements. Nevertheless, full MC simulations to estimate scatter and MC-SSS might be interesting to explore their capacity to further reduce bias that is still observed in HRRT reference tissue model studies. In the study presented in chapter 4 increased bias and variation of data were seen when SUVmax was used for regional assessment. The use of the maximum uptake value as a metric tends to have larger biases for the bigger spheres, both in TOF and TOF+PSF reconstructions. This findings indicate that more accurate and reliable tumour uptake quantification metrics are needed for studies in cancer patients. A drawback of this study is that only data from a single scanner were assessed, therefore more data are needed to compare the results with those from other systems. In addition, different scanner systems, acquisition protocols, and reconstruction settings may yield more variability and thus it requires a complete assessment of variability using multiple repositioned phantom scans at different sites, with different scanners and clinically used protocols. [ 11C]phenytoin has been developed to measure the Pgp up-regulation in epilepsy subjects [34]. This thesis describes a complete pharmacokinetic assessment of [11C]phenytoin in order to find optimal kinetic analysis approach for such studies. The tracer shows promising results compared to other Pgp tracers such as (R)-[11C]verapamil and [11C]-N-desmethylloper-amide that are currently used in clinical studies [35, 36]. We found that [11C]phenytoin provides a stable metabolism during the whole duration of the scan that is comparable between subjects. This provides the possibility for either the use of a plasma input function without additional metabolite analysis, or use of a population based correction for tracer metabolites. The tracer 113 General discussions and future perspectives Summary, general discussion and future perspectives has a relatively high VT in the brain with a low efflux (k2) values when compared to other tracers such as (R)-[11C]verapamil and [11C]-N-desmethylloperamide. Based on the findings in this thesis, it can be concluded that study time can be reduced without a notable loss of quantitative accuracy. The parametric methods showed pharmacokinetic values comparable to NLR. However, a future study should be performed to first validate the parametric methods for patients with epilepsy and subsequently compare the pharmacokinetic values between epilepsy patients and healthy subjects to investigate the clinical applicability of the tracer. 114 Bibliography [1] Turkington, T.G. PET Imaging Basics. Clinical PET-CT in Radiology. New York: Springer; 2010: p 21-8. [2] Kessler RM, Ellis JR, Jr., Eden M (1984) Analysis of emission tomographic scan data: limitations imposed by resolution and background. J Comput. Assist. Tomogr. 8: 514- 522 [3] Hudson HM, Larkin RS (1994) Accelerated image reconstruction using ordered subsets of projection data. IEEE Trans. Med Imaging 13: 601-609 [4] Comtat C, Kinahan PE, Defrise M, Michel C, Townsend DW (1997) Fast reconstruction of 3D PET data with accurate statistical modeling. IEEE Nuclear Science Symposium Conference 2: 963-967 [5] Nuyts J, Michel C, Dupont P (2001) Maximum-likelihood expectation-maximization reconstruction of sinograms with arbitrary noise distribution using NEC-transformations. IEEE Trans. Med Imaging 20: 365-375 [6] Liow JS, Strother SC, Rehm K, Rottenberg DA (1997) Improved resolution for PET volume imaging through three-dimensional iterative reconstruction. J Nucl Med 38: 1623-1631 [7] Lewellen TK (1998) Time-of-flight PET. Semin. Nucl Med 28: 268-275 [8] Muehllehner G, Karp JS (2006) Positron emission tomography. Phys. Med Biol. 51: R117-R137 [9] Conti M (2011) Focus on time-of-flight PET: the benefits of improved time resolution. Eur J Nucl Med Mol Imaging 38: 1147-1157 [10] Armstrong IS, James JM, Williams HA, Kelly MD, Matthews JC (2015) The assessment of time-of-flight on image quality and quantification with reduced administered activity and scan times in 18F-FDG PET. Nucl Med Commun. 36: 728-737 [11] Surti S, Scheuermann J, El FG, et al. (2011) Impact of time-of-flight PET on wholebody oncologic studies: a human observer lesion detection and localization study. J Nucl Med 52: 712-719 [12] Schaefferkoetter JD, Yan J, Townsend DW, Conti M (2015) Initial assessment of image 115 BIBLIOGRAPHY BIBLIOGRAPHY quality for low-dose PET: evaluation of lesion detectability. Phys. Med Biol. 60: 5543- 5556 [13] Akamatsu G, Mitsumoto K, Taniguchi T, Tsutsui Y, Baba S, Sasaki M (2014) Influences of point-spread function and time-of-flight reconstructions on standardized uptake value of lymph node metastases in FDG-PET. Eur J Radiol. 83: 226-230 [14] Shao X, Shao X, Wang X, Wang Y (2016) Applications of both time of flight and point spread function in brain PET image reconstruction. Nucl Med Commun. 37: 422-427 [15] Tong S, Alessio AM, Thielemans K, Stearns C, Ross S, Kinahan PE (2010) Properties of edge artifacts in PSF-based PET reconstruction. IEEE Nuclear Science Symposium Conference 3649-3652 [16] Rahmim A, Qi J, Sossi V (2013) Resolution modeling in PET imaging: theory, practice, benefits, and pitfalls. Med Phys. 40: 064301 [17] Brasse D, Kinahan PE, Lartizien C, Comtat C, Casey M, Michel C (2005) Correction methods for random coincidences in fully 3D whole-body PET: impact on data and image quality. J Nucl Med 46: 859-867 [18] Bendriem B, Trebossen R, Frouin V, Syrota A (1993) A PET scatter correction using simultaneous acquisitions with low and high lower energy thresholds. IEEE Nuclear Science Symposium Conference 1779-1783 [19] Grootoonk S, Spinks TJ, Sashin D, Spyrou NM, Jones T (1996) Correction for scatter in 3D brain PET using a dual energy window method. Phys. Med Biol. 41: 2757-2774 [20] Alexander Werling OBJDL-EA, Gunnar Brix (2002) Fast implementation of the single scatter simulation algorithm and its use in iterative image reconstruction of PET data. Physics in Medicine and Biology 47: [21] Watson CC (2000) New, faster, image-based scatter correction for 3D PET. IEEE Nuclear Science Symposium Conference 47: 1587-1594 [22] Watson C (2007) Extension of Single Scatter Simulation to Scatter Correction of Timeof-Flight PET. IEEE Nuclear Science Symposium Conference 54: 1679-1686 [23] Defrise M, Rezaei A, Nuyts J (2014) Transmission-less attenuation correction in timeof-flight PET: analysis of a discrete iterative algorithm. Phys. Med Biol. 59: 1073-1095 [24] Kinahan PE, Townsend DW, Beyer T, Sashin D (1998) Attenuation correction for a combined 3D PET/CT scanner. Med Phys.25: 2046-2053 [25] Kinahan PE, Hasegawa BH, Beyer T (2003) X-ray-based attenuation correction for positron emission tomography/computed tomography scanners. Semin. Nucl Med 33: 166-179 [26] Cheebsumon P, Boellaard R, de Ruysscher D, et al. (2012) Assessment of tumour size in PET/CT lung cancer studies: PET- and CT-based methods compared to pathology. EJNMMI Res. 2: 56 [27] Svarer C, Madsen K, Hasselbalch SG, et al. (2005) MR-based automatic delineation of 116 BIBLIOGRAPHY BIBLIOGRAPHY volumes of interest in human brain PET images using probability maps. Neuroimage. 24: 969-979 [28] Hammers A, Koepp MJ, Free SL, et al. (2002) Implementation and application of a brain template for multiple volumes of interest. Hum. Brain Mapp. 15: 165-174 [29] Akaike H (1974) A new look at the statistical model identification. Automatic Control, IEEE Transactions on 19: 716-723 [30] Logan J, Fowler JS, Volkow ND, et al. (1990) Graphical analysis of reversible radioligand binding from time-activity measurements applied to [N-11C-methyl]-(-)-cocaine PET studies in human subjects. J Cereb. Blood Flow Metab 10: 740-747 [31] Patlak CS, Blasberg RG, Fenstermacher JD (1983) Graphical evaluation of blood-tobrain transfer constants from multiple-time uptake data. J Cereb. Blood Flow Metab 3: 1-7 [32] Gunn RN, Gunn SR, Turkheimer FE, Aston JA, Cunningham VJ (2002) Positron emission tomography compartmental models: a basis pursuit strategy for kinetic modeling. J Cereb. Blood Flow Metab 22: 1425-1439 [33] Rezaei A, Michel C, Casey ME, Nuyts J (2016) Simultaneous reconstruction of the activity image and registration of the CT image in TOF-PET. Phys. Med Biol. 61: 1852-1874 [34] Verbeek J, Eriksson J, Syvanen S, et al. (2012) [11C]phenytoin revisited: synthesis by [11C]CO carbonylation and first evaluation as a Pgp tracer in rats. EJNMMI Res. 2: 36 [35] Lubberink M, Luurtsema G, van Berckel BN, et al. (2007) Evaluation of tracer kinetic models for quantification of P-glycoprotein function using (R)-[11C]verapamil and PET. J Cereb. Blood Flow Metab 27: 424-433 [36] Kreisl WC, Liow JS, Kimura N, et al. (2010) P-glycoprotein function at the blood-brain barrier in humans can be quantified with the substrate radiotracer 11C-N-desmethylloperamide. J Nucl Med 51: 559-566 117