Respiratory gated PET/CT in a European multicentre retrospective study: added diagnostic value in detection and characterization of lung lesions

Eur J Nucl Med Mol Imaging (2012) 39:1381 1390 DOI 10.1007/s00259-012-2148-2 ORIGINAL ARTICLE Respiratory gated PET/CT in a European multicentre retrospective study: added diagnostic value in detection and characterization of lung lesions Luca Guerra & Elena De Ponti & Federica Elisei & Valentino Bettinardi & Claudio Landoni & Maria Picchio & Maria Carla Gilardi & Annibale Versari & Federica Fioroni & Miroslaw Dziuk & Magdalena Koza & Renée Ahond-Vionnet & Bertrand Collin & Cristina Messa Received: 1 February 2012 /Accepted: 24 April 2012 /Published online: 16 May 2012 # Springer-Verlag 2012 Abstract Purpose The aim of our work is to evaluate the added diagnostic value of respiratory gated (4-D) positron emission tomography/computed tomography (PET/CT) in lung lesion detection/characterization in a large patient population of a multicentre retrospective study. Methods The data of 155 patients (89 men, 66 women, mean age 63.9±11.1 years) from 5 European centres and submitted to standard (3-D) and 4-D PET/CT were retrospectively analysed. Overall, 206 lung lesions were considered for the analysis (mean ± SD lesions dimension 14.7 ± 11.8 mm). Maximum standardized uptake values (SUV max ) and lesion detectability were assessed for both 3-D and 4-D PET/CT studies; 3-D and 4-D PET/CT findings were compared to clinical follow-up as standard reference. Results Mean ± SD 3-D and 4-D SUV max values were 5.2 ± 5.1 and 6.8 ± 6.1 (p<0.0001), respectively, with an average percentage increase of 30.8 %. In 3-D PET/CT, 86 of 206 L. Guerra (*) : F. Elisei : C. Messa Nuclear Medicine, San Gerardo Hospital, Via Pergolesi 33, 20900 Monza, Italy e-mail: l.guerra@hsgerardo.org E. De Ponti Medical Physics, San Gerardo Hospital, Monza, Italy V. Bettinardi : C. Landoni : M. Picchio : M. C. Gilardi Nuclear Medicine, San Raffaele Scientific Institute, Milan, Italy V. Bettinardi : M. Picchio : M. C. Gilardi : C. Messa Institute for Bioimaging and Molecular Physiology, National Research Council, Milan, Italy C. Landoni University of Milano-Bicocca, Milan, Italy M. C. Gilardi : C. Messa Tecnomed Foundation, University of Milano-Bicocca, Milan, Italy A. Versari Nuclear Medicine, Scientific Institute Santa Maria Nuova Hospital, Reggio Emilia, Italy F. Fioroni Medical Physics, Scientific Institute Santa Maria Nuova Hospital, Reggio Emilia, Italy M. Dziuk Department of Nuclear Medicine, Military Institute of Medicine, Masovian PET-CT Centre, Warsaw, Poland M. Koza Masovian PET-CT Centre, Euromedic Diagnostic, Warsaw, Poland R. Ahond-Vionnet : B. Collin Service de Médecine Nucléaire, Hôpital Pierre Beregovoy, Nevers Cedex, France

1382 Eur J Nucl Med Mol Imaging (2012) 39:1381 1390 (41.7 %) lesions were considered positive, 70 of 206 (34 %) negative and 50 of 206 (24.3 %) equivocal, while in 4-D PET/CT 117 of 206 (56.8 %) lesions were defined as positive, 80 of 206 (38.8 %) negative and 9 of 206 (4.4 %) equivocal. In 34 of 50 (68 %) 3-D equivocal lesions followup data were available and the presence of malignancy was confirmed in 21 of 34 (61.8 %) lesions, while in 13 of 34 (38.2 %) was excluded. In 31 of these 34 controlled lesions, 20 of 34 (58.8 %) and 11 of 34 (32.4 %) were correctly classified by 4-D PET/CT as positive and negative, respectively; 3 of 34 (8.8 %) remained equivocal. With equivocal lesions classified as positive, the overall accuracy of 3-D and 4-D was 85.7 and 92.8 %, respectively, while the same figures were 80.5 and 94.2 % when equivocal lesions were classified as negative. Conclusion The respiratory gated PET/CT technique is a valuable clinical tool in diagnosing lung lesions, improving quantification and confidence in reporting, reducing 3-D undetermined findings and increasing the overall accuracy in lung lesion detection and characterization. Keywords Respiratory gating. Positron emission tomography. Computed tomography. Lung lesion detection/characterization Introduction Respiratory movement is a relevant factor affecting PET/CT image quality and quantitative accuracy of lung lesions. In fact, due to respiratory movement a lesion may present distortion of boundaries and volume in CT images, while in PET images, it can appear blurred and thus overestimated in its metabolic volume as well as underestimated in its maximum standardized uptake value (SUV max )[1]. In addition, in a PET/CT procedure, the different acquisition time of a CT scan (usually a few seconds) from that of a PET scan (min/bed) can lead to a misalignment between CT and PET images, resulting in an incorrect attenuation correction map and consequently incorrect quantitative (e.g. SUV max )values [2 4]. The degradation of image quality in lung lesions due to the respiratory movement may reduce the sensitivity of the imaging technique, particularly for small lesions typically more affected by the partial volume effect [5] and for those located in the basal part of the lungs, where the respiratory motion is more pronounced [6]. In fact, although a very high overall sensitivity of 18 F-fluorodeoxyglucose (FDG) PET in detecting lung nodules has been demonstrated (96.8 %) [7], this figure decreases consistently (70 %) for lesions smaller than 2 cm [7 9]. In recent years, technological tools aimed at improving PET/CT image quality and quantification have been developed. The technique of respiratory gating in PET/CT (4-D PET/CT), able to synchronize PET and CT acquisition to respiratory motion, may represent a valid technology for improving image quality of lung lesions. In particular, the reduction/elimination of the artefacts related to respiratory movement both in CT and PET demonstrated an improvement in quantitative data, with a percentage increase of SUV max values in a range from 22 to 83 % [10 12], a reduction of metabolic volume of the lesion of 44.5 % [11] and an increased detection rate of malignant lesions in comparison to standard PET/CT acquisition [12]. The possible diagnostic role of 4-D PET/CT in lung lesions has also been assessed, but in a small number of patients [12], suggesting its application in a larger population. The aim of our work is to evaluate the possible added diagnostic value of 4-D PET/CT in lung lesion detection/ characterization in a large patient population of a multicentre retrospective study. Materials and methods Patient population The data of 155 patients (89 men, 66 women, mean age 63.9 ±11.1 years) from 5 European centres (San Raffaele Scientific Institute in Milan, Italy, San Gerardo Hospital in Monza, Italy, Santa Maria Nuova Hospital in Reggio Emilia, Italy, Centre Hospitalier Nevers in Nevers, France, Masovian PET/CT Centre in Warsaw, Poland) and submitted to 3- D and 4-D PET/CT from February 2007 to December 2010 were retrospectively analysed. Overall, 206 lung lesions were considered for the analysis. The mean ± SD lesion dimension was 14.7 ± 11.8 mm at CT (range 2 80 mm). In 142 of 155 (91.6 %) patients PET/CT scans were performed for restaging oncological disease [10 patients (6.4 %) with breast cancer, 31 (20 %) with gastrointestinal cancer, 66 (42.6 %) with lung cancer and 35 (22.6 %) with other tumours] and in 13 of 155 (8.4 %) patients for solitary pulmonary nodule characterization. Patient preparation Patients were prepared for PET/CT examination according to common guidelines [13]. In particular, the patient was kept fasting at least 6 h before tracer administration, blood glucose was assessed immediately before tracer administration and oral hydration was administered in the waiting time before scanning. The 18 F-FDG administered dose was 3.7 4.4 MBq/kg and the scan started 60 90 min after the administration of the tracer. Patients were asked to void their bladder immediately before scanning. No specific patient preparation was needed in view of the 4-D PET/CT

Eur J Nucl Med Mol Imaging (2012) 39:1381 1390 1383 procedure besides detailed information explaining the aim of the 4-D study and the importance of an active collaboration of the patient in maintaining a regular breathing pattern for the entire time of the study. In this respect, after having positioned the patient on the scanner s bed and immediately before the beginning of the PET/CT procedure, a simple training session of 1 2 min was performed recording the breathing curve of the patient to evaluate his/her ability to keep the breathing regular. In case of irregular breathing, the operator verbally stimulated the patient, sometimes with coaching instructions (e.g. breath in, breath out), to help him/her in keeping a regular and relaxed breathing condition. Patient preparation protocols are reported in Table 1. Whole-body (WB) 3-D PET/CT acquisition and 4-D PET/CT PET/CT scans were performed with Discovery STE system (GE Healthcare, Milwaukee, WI, USA) in three centres, Discovery 600 (GE Healthcare, Milwaukee, WI, USA) in one centre and Discovery RX system (GE Healthcare, Milwaukee, WI, USA) in one centre. Two types of acquisition protocols were used in the different centres: a conventional WB PET/CT procedure followed by a 4-D PET/CT study or an integrated WB and 4-D PET/CT protocol. In case of a conventional WB PET/CT procedure followed by a 4-D PET/CT study, the examination session consisted in a scout view to define the scan volume, including the skull base and the upper thighs both for CT and PET, followed by a WB CT scan (helical mode) and a WB PET scan (3-D frame mode) corresponding to a mean of 6 7 PET fields of view (FOVs). Once the WB PET/CT study terminated, the 4-D PET/CT protocol consisted in a 4-D CT scan (cine mode) over the anatomical region of interest. The 4-D CT study was performed by using a step-and-shoot cine mode acquisition technique. In the CT cine modality, the X- ray tube rotates around the patient in the same axial position for a time lasting at least one full patient breathing cycle, after which the X-ray beam is automatically turned off and the table is moved into the next adjacent axial (Z) position. This process is repeated until the whole anatomical region of interest is fully covered. The cine duration parameter to be set for the acquisition (corresponding to the mean of the patient breathing period) was measured by a respiratory monitoring system. After termination of the 4-D CT scan, a 4-D PET scan (3-D list mode) of one or two PET FOVs (8, 9, 10 or 12 min/fov) centred over the same anatomical region of interest was acquired. The respiratory monitoring system used for both 4-D PET and 4-D CT scans was the Real Time Position Management system (RPM, Varian Oncology Systems, Palo Alto, CA, USA) [14, 15]. The RPM provides the synchronization between the patient s respiratory cycle and data acquisition. In particular, the RPM tracks the vertical displacement of two infrared reflective markers attached on a small plastic box positioned on the thorax or upper abdomen of the patient. During the 4-D CT acquisition, the RPM system records and synchronizes the spatial position of the markers and all the specific information related to the CT scan (time, Z position, X-ray on/off etc.). During the 4-D PET scan, the RPM system provides a trigger to the PET system at a specific moment (e.g. end of inspiration) for each of the patient s breathing cycles. In case of an integrated WB and 4-D PET/CT study the examination session consisted in a scout view for the scan volume definition (6 7 PET FOVs) followed by a WB CT scan (helical mode), a 4-D CT scan (cine mode) over the anatomical region of interest and a WB PET scan (3-D list mode). In this case, one or two of the PET FOVs centred over the anatomical region and covered by 4-D CT were acquired for 8, 9, 10 or 12 min with the synchronization to the patient s respiratory signal (4-D PET). Specific protocol and acquisition parameters used in each PET/CT protocol in the different centres are reported in Table 1. 3-D and 4-D PET/CT data processing and reconstruction For each conventional WB PET/CT study a set of PET images (3-D WB ) was reconstructed with a 3-D ordered subset expectation maximization (OSEM) algorithm, over a 128 or 256 pixels matrix, with attenuation, random and scatter correction. Attenuation correction of PET data was performed by using the corresponding helical CT images. The 4-D PET data processing consisted in the sorting of the acquired raw data by using a proportional phase sorting technique. Four, five or six 4-D PET phases were generated depending on the acquisition time of the 4-D PET study. In particular, for 8, 9, 10 and 12 min of 4-D PET acquisition, 4, 6, 5 and 6 phases were generated, respectively (Table 1). The 4-D PET phase data were then reconstructed by using a 3-D OSEM algorithm over a matrix size of 128 or 256 pixels, with attenuation, random and scatter correction. In this case, attenuation correction was performed for each 4-D PET phase by using the corresponding 4-D CT phase (phase-matched reconstruction). The 4-D PET raw data were also used to generate a second set of conventional 3-D PET data set (degraded by the respiratory motion) with the same duration of the corresponding 4-D PET phases (e.g. 1.5 or 2 min). These procedure data were possible only for 102 of 155 (66 %) patients as for the remaining 53 of 155 (34 %) patients (70/ 206 lesions, 34 %) 4-D PET raw data were no longer available for retrospective data processing. This new set of 3-D PET data (3-D U4D ) was then reconstructed with the same parameters used for the reconstruction

1384 Eur J Nucl Med Mol Imaging (2012) 39:1381 1390 Table 1 Patient s preparation protocol and 3-D and 4-D PET/CT acquisition and processing parameters in participating centres Reggio Emilia Warsaw Milan Monza Nevers Patient preparation No. of patients studied 50 36 24 36 9 No. of lesions 64 36 58 39 9 Enrolment time (years) 2006 2010 2008 2010 2008 2010 2009 2010 2010 Fasting time 6 h 6 h 6 h 6 h 6 h Blood glucose level threshold 200 mg/dl 150 mg/dl 200 mg/dl 170 mg/dl 180 mg/dl Dose administered 3.7 MBq/kg 4.4 MBq/kg 3.7 MBq/kg 3.7 MBq/kg 3.7 MBq/kg Uptake time 60 90 min 60 80 min > 60 min 60 80 min 60 80 min PET/CT scanner Discovery STE Discovery STE Discovery STE Discovery 600 Discovery RX Static PET/CT acquisition protocol CT modality Helical Helical Helical Helical Helical ma Fixed: 80 ma Auto ma range 10 120; noise index 22 Auto ma range 30 150; noise index 25 Fixed: 40 ma Auto ma range 30 120; noise index 20 kv 120 kv 120 kv 140 kv 120 kv 120 kv Rotation time 0.8 s 0.6 s 0.8 s 0.8 s 0.8 s Slice thickness 3.75 mm 3.75 mm 3.75 mm 3.75 mm 3.75 mm Pitch 1.375 1.375 1.375 1.375 1.75 PET min/fov 3:30 2:20 2:30 2:00 1:30 Matrix size 256 256 128 128 128 128 256 256 128 128 4-D PET/TC acquisition protocol CT modality Step and shoot Step and shoot Step and shoot Step and shoot Step and shoot ma Fixed: 80 ma Fixed: 50 ma Fixed: 40 ma Fixed: 40 ma Fixed: 30 ma kv 120 kv 120 kv 120 kv 120 kv 120 kv Rotation time 0.5 s 0.5 s 0.5 s 0.5 s 0.5 s Revolution duration Breathing period+0.5 s Breathing period+0.5 s Breathing period+0.5 s Breathing period+0.5 s Breathing period+0.5 s Slice thickness 2.5 mm 2.5 mm 2.5 mm 2.5 mm 2.5 mm PET min/phase 2 min 2 min 2 min 2 min 1:30 min No. of phases 6 or 4 5 4 6 6 Matrix size 256 256 128 128 128 128 256 256 128 128 PET reconstruction protocol Reconstruction technique OSEM OSEM OSEM OSEM OSEM No. of of iterations 2 2 2 2 2 No. of subsets 28 28 28 28 21 Postfilter 5 mm 6 mm 5.1 mm 5 mm 6 mm FOV field of view, OSEM ordered subset expectation maximization of the 4-D PET phases. In such a case the attenuation correction of the PET data was performed by using an average CT obtained by the 4-D CT images. The 4-D PET/CT reconstruction parameters are reported in Table 1. Image analysis All 3-D and 4-D PET/CT reconstructed images were displayed in transaxial, coronal and sagittal sections. Depending on the working modality of each centre, image analysis was performed by one or two expert nuclear medicine physicians aware of the patient s clinical data at the time of PET/CT scan execution. The 3-D and 4-D PET/CT images were reviewed by the readers for defining the target lesion as positive, equivocal or negative for neoplastic disease according to visual criteria. In particular, the lesion s activity was compared to mediastinal blood pool background activity and was considered positive if lesion uptake was visually significantly higher than mediastinal background, negative if no significant visible uptake was

Eur J Nucl Med Mol Imaging (2012) 39:1381 1390 1385 present and equivocal if neither positive nor negative criteria were matched. Quantitative analysis Quantitative analysis was performed by comparing the SUV max values as calculated on 3-D U4D images (66 % of the patients) or 3-D WB (34 % of the patients) and the corresponding 4-D PET phases. For each lesion with any visible uptake the 3-D and 4-D PET SUV max was recorded and stored in a database. For all respiratory gated studies, the 4-D PET phase with the highest SUV max (best bin) was used for quantitative evaluation. SUV max values were not recorded if no visible uptake was detected by the reader or if the SUV max value was lower than 1.0. In these cases the lesion was classified as no uptake. Standard reference In order to calculate overall diagnostic accuracy of both imaging techniques, 3-D and 4-D PET/CT results were compared to clinical follow-up data, including biopsies, diagnostic imaging (MRI, contrast-enhanced CT, ultrasounds studies), physical examination, laboratory tests and medical records. Compared to standard reference, 3-D and 4-D PET/CT results were classified as true-positive (TP), false-positive (FP), true-negative (TN) and false-negative (FN). As equivocal lesions cannot be considered either positive or negative, they were included in the analysis twice, considering them once as positive and once as negative for neoplastic disease. Sensitivity, specificity and accuracy of both imaging techniques were then calculated with and without including equivocal lesions. Statistical analysis The paired Student s t test was used to compare 3-D and 4-D PET/CT SUV max values. Absolute and percentage frequency were calculated for the assessment of concordance between 3-D and 4-D findings. Sensitivity, specificity and accuracy were calculated to assess diagnostic performance for imaging techniques. Stata software 9.0 (1999, StataCorp, College Station, TX, USA) was used for statistical analysis. A level of p<0.05 was adopted for significance. Results Quantitative analysis Of 206 (36.4 %) lesions, 75 were considered as having no uptake in 3-D PET/CT. The mean ± SD dimension of these lesions was 9.1 ± 4.9 mm (range 2 40). Of the remaining 131 lesions, 2 of them were not included in the analysis; 1 lesion became negative in the 4-D study and in another one SUV max was not calculated because of missing PET data (patient weight and administered activity). In 129 lesions (mean dimension ± SD 17.4 ± 12.2 mm, range 4 80) 3-D and 4-D SUV max values were 5.2 ± 5.1 and 6.8 ± 6.1 (p< 0.0001), respectively. These lesions were divided according to dimension measured on CT in group 1 (n041; up to 10 mm), group 2 (n036; 11 15 mm) and group 3 (n052; larger than 15 mm). In groups 1, 2 and 3, the 3-D and 4-D mean ± SD SUV max values were 2.9 ± 1.7 and 3.7 ± 2.1 (p< 0.0001), 4.1 ± 2.7 and 5.9 ± 3.7 (p<0.0001), and 7.8 ± 6.8 and 9.8 ± 7.9 (p<0.0001), respectively. The quantitative analysis results are summarized in Table 2. Visual analysis Overall, in 3-D PET/CT, 86 of 206 (41.7 %) lesions were considered positive, 70 of 206 (34 %) negative and 50 of 206 (24.3 %) equivocal, while in 4-D PET/CT 117 of 206 (56.7 %) lesions were defined as positive, 80 of 206 (38.8 %) negative and 9 of 206 (4.4 %) equivocal. The results of 3-D and 4-D PET/CT were discordant in 51 of 206 (24.7 %) lesions. Particularly, 1 of 86 (1.2 %) positive lesions at 3-D PET/CT became negative at 4-D study and 1 of 86 (1.2 %) equivocal. Of 70 (4.3 %) negative lesions at 3- D PET/CT, 3 became positive at 4-D study and 2 of 70 (2.8 %) were considered equivocal. Finally, 30 of 50 (60 %) equivocal lesions at 3-D PET/CT became positive with the 4- D technique and 14 of 50 (28 %) were considered negative. Visual analysis results are summarized in Table 3. In 191 of 206 lesions, dimensional data were available for the analysis; as previously mentioned, three groups were considered according to the dimensional criteria indicated above. In group 1, 12 of 24 (50 %) equivocal lesions at 3-D PET/CT became positive at 4-D PET/CT study, while 9 of 24 (37.5 %) were considered negative. In group 2, 11 of 14 (78.6 %) equivocal lesions in 3-D scan became positive and 2 of 14 (14.3 %) negative at 4-D PET/CT scan. In group 3, 7 Table 2 SUV max analysis results Lesion diameter (n) Mean SUV max ±SD 3-D PET/CT 4-D PET/CT p (Student s t test) All lesions 5.2 ± 5.1 6.8 ± 6.1 < 0.0001 (n0129) Group 1 (n041) 2.9 ± 1.7 3.7 ± 2.1 < 0.0001 Group 2 (n036) 4.1 ± 2.7 5.9 ± 3.7 < 0.0001 Group 3 (n052) 7.8 ± 6.8 9.8 ± 7.9 < 0.0001

1386 Eur J Nucl Med Mol Imaging (2012) 39:1381 1390 Table 3 Comparison of visual analysis results for 3-D and 4-D PET/CT (all lesions), expressed as absolute value and row percentage of total lesions 4-D PET/CT Positive Negative Equivocal Total 3-D PET/CT Positive 84 (97.6 %) 1 (1.2 %) 1 (1.2 %) 86 Negative 3 (4.4 %) 65 (92.8 %) 2 (2.8) 70 Equivocal 30 (60.0 %) 14 (28 %) 6 (12.0 %) 50 Total 117 80 9 206 of 10 (70 %) equivocal lesions at 3-D PET/CT changed to positive and 1 of 10 (10 %) to negative at 4-D PET/CT. Visual analysis results in grouped lesions are reported in Table 4. Comparison of 3-D and 4-D PET/CT results to standard reference Follow-up data for validation of 3-D and 4-D PET/CT results were available in 154 of 206 (74.7 %) lesions. According to follow-up data, 87 of 154 (56.5 %) and 67 of 154 (43.5 %) lesions were definitely considered neoplastic and benign, respectively. In 3-D PET/CT, 69 of 154 (44.8 %) lesions were defined as positive, 51 of 154 (33.1 %) negative and 34 of 154 (22.1 %) equivocal. Excluding the equivocal lesions from the analysis, sensitivity, specificity and accuracy were 95.4, 88.9 and 92.5 %, respectively. According to follow-up data, thepresenceofmalignancywasconfirmedin21of34 (61.8 %) equivocal lesions, while in 13 of 34 (38.2 %) was excluded. Considering 3-D equivocal lesions as positive, sensitivity, specificity and accuracy were 96.6, 71.6 and 85.7 %, respectively, while the same figures were 72.4, 91.0 80.5 % if 3-D equivocal findings were included as negative. In 4-D PET/CT 90 of 154 (58.4 %) lesions were considered positive, 60 of 154 (39.0 %) negative and 4 of 154 (2.6 %) equivocal. In comparison to standard reference, sensitivity, specificity and accuracy were 98.8, 90.8 and 95.3 %, respectively. According to follow-up data, neoplastic disease was confirmed in 2 of 4 (50 %) equivocal lesions, while another 2 of 4 (50 %) were definitely assessed as benign. With equivocal lesions as positive, sensitivity, specificity and accuracy were 98.9, 88.1 and 94.2 %, respectively; the same values became 96.6, 91.0 and 94.2 % when equivocal lesions were considered as negative. The overal results for 3-D and 4-D PET/CT are reported in Table 5. Two clinical cases of 3-D and 4-D PET/CT results are illustrated in Figs. 1 and 2. Discussion Respiratory movement is one of the most important factors affecting PET/CT diagnostic performance. This is mainly due to a distribution of the lesion s activity in a volume larger than the real lesion s volume and to an incorrect attenuation correction map caused by a mismatch between CT and PET images. It is reasonable to suppose that the control of respiratory movement could be of great benefit both for quantification and lesion detection due to the increase of the lesion s metabolic signal. In this work we compared the SUV max of the lesions as calculated in standard 3-D PET images (degraded by respiratory motion) and in the corresponding respiratory gated 4-D PET images. We found that the mean 4-D PET SUV max value (6.8±6.1) was significantly higher than the corresponding 3-D value (5.2±5.1; p<0.0001) with a mean increase of 30.8 %. This result is partially in agreement with previously published [10 12] works in which a higher SUV max increase was found. Table 4 Comparison of visual analysis results for 3-D and 4-D PET/CT for lesion dimensions in three groups. Both absolute and row percentage values of total lesions are reported 4-D PET/CT Group 1 Group 2 Group 3 Pos. Neg. Eq. Total Pos. Neg. Eq. Total Pos. Neg. Eq. Total 3-D PET/CT Pos. 22 (96 %) 0 1 (4 %) 23 23 (100) 0 0 23 39 (100 %) 0 0 39 Neg. 1 (2 %) 41 (95 %) 1 (2 %) 43 1 (13 %) 7 (88 %) 0 8 1 (14 %) 5 (71 %) 1 (14 %) 7 Eq. 12 (50 %) 9 (38 %) 3 (13) 24 11 (79 %) 2 (14 %) 1 (7 %) 14 7 (70 %) 1 (10 %) 2 (20 %) 10 Total 35 50 5 90 35 9 1 45 47 6 3 56

Eur J Nucl Med Mol Imaging (2012) 39:1381 1390 1387 Table 5 3-D and 4-D PET/CT diagnostic results TP TN FP FN Sens. Spec. Acc. 3-D PET/CT No equivocal lesions 63/69 (91.3 %) 48/51 (94.1 %) 6/69 (8.7 %) 3/51 (5.9 %) 95.4 % 88.9 % 92.5 % Equivocal lesions as positive 84/103 (81.6 %) 48/51 (94.1 %) 19/103 (18.4 %) 3/51 (5.9 %) 96.6 % 71.6 % 85.7 % Equivocal lesions as negative 63/69 (91.3 %) 61/85 (71.8 %) 6/69 (8.7 %) 24/85 (28.2 %) 72.4 % 91.0 % 80.5 % 4-D PET/CT No equivocal lesions 84/90 (93.3 %) 59/60 (98.3 %) 6/90 (6.7 %) 1/60 (1.7 %) 98.8 % 90.8 % 95.3 % Equivocal lesions as positive 86/94 (91.5 %) 59/60 (98.3 %) 8/94 (8.5 %) 1/60 (1.7 %) 96.5 % 88.0 % 92.8 % Equivocal lesions as negative 84/90 (93.3 %) 61/64 (95.3 %) 6/90 (6.7 %) 3/64 (4.7 %) 96.6 % 91.0 % 94.2 % In particular, Lupi et al. [10] found an increase of the metabolic parameter of 77.2 %, while García Vicente et al. [12] found an increase of 83.3 %. These percentages are considerably higher than those found in our population and could be due to various factors, such as the different characteristics of the patients (type of tumours, dimension of the lesion, lesion location, lesion displacement), type of PET acquisition (2-D or 3-D) [10] and also to the longer uptake time of 4-D scans with respect to that of the corresponding 3-D study [12]. In fact, SUV in neoplastic lung lesions constantly increases with time and higher values are expected in delayed acquisition protocols [16]. In our work, for most of the patients (66 %) 3-D and 4-D SUV max values were obtained from data sets acquired at the same time and processed to have the same acquisition time; thus, the differences found in quantification were mainly due to physical effect than to tracer kinetics. In 34 % of the cases 3-D and 4-D images could not be obtained by reprocessing the same data set; thus, in such cases a possible bias due to differences in time and statistics could have affected such a percentage of the data. On the other hand, the time difference between WB 3-D PET and 4-D PET was about 15 20 min, a time in which the tracer uptake should not increase considerably. Differences in count statistics between 3-D and 4-D acquisitions (e.g. 3.5 min versus 2 min and 2.5 min versus 2 min in two centres) are more difficult to be evaluated as the two acquisitions result in a different distribution of the activity. In fact, in the case of two Fig. 1 A male patient, 63 years old, heavy smoker, underwent 4-D PET/CT integrated in the WB scan for characterizing a right lung nodule with 12 mm diameter at CT (b: black arrow). a Maximum intensity projection image. c Fusion image. In 3-D PET (d) a very faint uptake was visible (black arrow; SUV max 1.5) and the nodule was scored as negative for disease. In the best bin 4-D PET image (e) the tracer uptake significantly increased (black arrow; SUVmax 2.8) and the lesion was considered positive for disease. On the basis of the 4-D finding, the patient underwent surgery and an adenocarcinoma was diagnosed at histological examination

1388 Eur J Nucl Med Mol Imaging (2012) 39:1381 1390 Fig. 2 A female patient, 53 years old, underwent 4-D PET/CT integrated in the WB scan for characterization of a nodule (20 mm diameter) in the right lung lower lobe at CT (b: black arrow). a Maximum intensity projection image. c Fusion image. In 3-D PET (d) 18 F-FDG uptake was very low (black arrow; SUV max 1.3) and the nodule was scored as probably not malignant. In the best bin 4-D image (e) there was an increase of the uptake (black arrow; SUV max 3.2) and the report changed to malignant lesion. The patient underwent surgery and the histological finding was non-small cell lung carcinoma acquisitions performed under the same conditions (e.g. a 3- D PET acquired during free breathing) but with different duration, the shorter one will have lower global as well as local statistics (e.g. in a moving lesion); thus, a comparison between the two images based on the SUV max values would be biased by a higher global as well as local noise in the shorter acquisition. On the other hand, when we compare a conventional 3-D acquisition with a 4-D PET phase sorted in a way to obtain the same acquisition time, global statistics will be the same or very similar but locally (e.g. moving lesion) the 4-D data will have less variations as the counts will be spatially more concentrated. Thus, a direct relationship between two different acquisition times in the case of a 3-D and a 4-D study is not easy to be established at a local level, as it depends on several factors (e.g. acquisition time, lesion displacement, lesion dimension). This consideration is also valid in relation to the choice of the 4-D PET phase with the highest SUV max (best bin) to be compared with the 3-D PET data. In fact, as the statistics of each bin is slightly different, depending on the residual amount of motion, it could be argued that the bin with the highest SUV max is not the best one but rather the bin with the lower statistics and thus with the higher noise. On the other hand, in a 4-D PETstudy, the best bin is expected to be the one with less residual amount of motion and thus the one with a more concentrated activity distribution. The 3-D and 4-D PET/CT SUV max values were also compared amongst lesions grouped according to dimension criteria. In all groups, the mean SUV max values of the 4-D study were significantly higher than those in static acquisition, with a percentage increase of 27.6, 43.9 and 25.6 % in groups 1, 2 and 3, respectively. These data indicate that changes in quantification due to motion correction are always present, even for larger lesions. According to visual analysis results, the main discrepancy between static and 4-D PET/CT data concerned equivocal findings. In particular, we found 50 of 206 (24.3 %) and only 9 of 206 (4.4 %) equivocal findings in 3-D and 4-D PET/CT scans, respectively. Most of the 3-D equivocal findings (30/50; 60.0 %) were shifted to definitely positive in 4-D PET/CT scans due to the increase of metabolic signal by motion control. Moreover, 14 of 50 lesions (28.0 %) were shifted to definitely negative, mainly because of a more precise location of lesions in the basal part of the right lung, correctly separating lung from liver activity. Finally and most importantly, only 9 of 206 (4.4 %) lesions remained equivocal at 4-D PET/CT scan. These results showed that the respiratory gating technique can make the physician more confident in the interpretation of PET images, reducing the number of undetermined findings that are usually unhelpful for clinical patient management.

Eur J Nucl Med Mol Imaging (2012) 39:1381 1390 1389 In 154 of 206 (74.7 %) lesions with follow-up data, 3-D and 4-D PET/CT scans showed excellent sensitivity, specificity and overall accuracy if only definite findings were considered. As equivocal findings are inconclusive from a clinical point of view, they were considered twice in the statistical analysis, once as positive and once as negative and compared to standard reference. With equivocal lesions as positive, 3-D and 4-D overall accuracy was 85.7 and 92.8 %, respectively, while the same figures were 80.5 and 94.2 % when equivocal lesions were considered as negative. The better accuracy of 4-D PET/CT is related to the added diagnostic value in the equivocal lesions shifted to positive or negative. In particular, 4-D scan correctly classified 3-D equivocal lesions in 31 of 50 (62 %) cases and was TP and TN in 20 of 50 (40 %) and 11 of 50 (22 %) cases, respectively. A similar improvement in diagnostic accuracy of the respiratory gating technique was also reported by García Vicente et al. [12] who demonstrated that 52 % of neoplastic lesions considered negative or equivocal at 3-D PET/CT were correctly classified at 4-D PET/CT. Although the clinical impact of 4-D results on patient management was not investigated in our study, the increase in diagnostic accuracy of the 4-D technique in lung lesion detection and characterization might be beneficial for the clinical decision-making in oncological patients either for excluding or confirming neoplastic lung lesions, as showed by clinical examples in Figs. 1 and 2. Although the 4-D technique should ideally be beneficial for any lung lesions, we have to consider that setting up the system and acquiring 4-D data require an additional time to the 3-D scan. In our study we found that 38 of 50 (76 %) equivocal lesions on 3-D scan were included in groups 1 and 2. In this context, the 4-D technique could be reserved for those patients with known small lung lesions to be characterized, or when unexpected/unknown equivocal lesions are found at 3-D study. As mentioned above, this is a retrospective multicentre study in which there could be some bias related to differences among the centres in the 4-D acquisition protocol (i.e. integrated versus sequential, number of phases), interpretation criteria of PET/CT findings and patient selection. These differences also reflect the difficulties in the implementation of the 4-D PET/CT technique in clinical routine, as it can be deduced from the difference in patient numbers and enrolment time among the centres (Table 1). This is mainly related to the differences in hardware and software devices, making the 4-D technique time consuming for the older PET/CT systems, usually not equipped with automatic protocol for 4-D image reconstruction. Nowadays the PET/CT scanners are implemented with fully automatic processes for 4-D data sorting and image reconstruction (GE Healthcare Discovery 600 system of one centre), making the 4-D technique feasible in a more straightforward and easy way in clinical routine. Notwithstanding these differences, to our knowledge, this is the first study assessing the diagnostic value of the 4-D PET/CT imaging technique in a large patient population, enrolled in five European nuclear medicine centres that utilize 4-D PET/CT scans for clinical purpose, with similar criteria for image acquisition, processing and analysis. In conclusion, the respiratory gated PET/CT technique is a valuable clinical tool in diagnosing lung lesions, able to improve quantification and confidence in reporting, to reduce 3-D undetermined findings and to increase the overall accuracy in lung lesion detection and characterization. Conflicts of interest References None. 1. Nehmeh SA, Erdi YE, Ling CC, Rosenzweig KE, Schoder H, Larson SM, et al. Effect of respiratory gating on quantifying PET images of lung cancer. J Nucl Med 2002;43:876 81. 2. Nehmeh SA, Erdi YE, Ling CC, Rosenzweig KE, Squire OD, Braban LE, et al. Effect of respiratory gating on reducing lung motion artifacts in PET imaging of lung cancer. Med Phys 2002;29:366 71. 3. Erdi YE, Nehmeh SA, Pan T, Pevsner A, Rosenzweig KE, Mageras G, et al. The CT motion quantitation of lung lesions and its impact on PET-measured SUVs. J Nucl Med 2004;45:1287 92. 4. Bettinardi V, Rapisarda E, Gilardi MC. Number of partitions (gates) needed to obtain motion-free images in a respiratory gated 4D-PET/CT study as a function of the lesion size and motion displacement. Med Phys 2009;36:5547 58. 5. Goerres GW, Kamel E, Seifert B, Burger C, Buck A, Hany TF, et al. Accuracy of image coregistration of pulmonary lesions in patients with non-small cell lung cancer using an integrated PET/ CT system. J Nucl Med 2002;43:1469 75. 6. Pevsner A, Nehmeh SA, Humm JL, Mageras GS, Erdi YE. Effect of motion on tracer activity determination in CT attenuation corrected PET images: a lung phantom study. Med Phys 2005;32:2358 62. 7. Gould MK, Maclean CC, Kuschner WG, Rydzak CE, Owens DK. Accuracy of positron emission tomography for diagnosis of pulmonary nodules and mass lesions: a meta-analysis. JAMA 2001;285:914 24. 8. Degirmenci B, Wilson D, Laymon CM, Becker C, Mason NS, Bencherif B, et al. Standardized uptake value-based evaluations of solitary pulmonary nodules using F-18 fluorodeoxyglucose-pet/ computed tomography. Nucl Med Commun 2008;29:614 22. 9. Kim SC, Machac J, Krynyckyi BR, Knesaurek K, Krellenstein D, Schultz B, et al. Fluoro-deoxy-glucose positron emission tomography for evaluation of indeterminate lung nodules: assigning a probability of malignancy may be preferable to binary readings. Ann Nucl Med 2008;22:165 70.

1390 Eur J Nucl Med Mol Imaging (2012) 39:1381 1390 10. Lupi A, Zaroccolo M, Salgarello M, Malfatti V, Zanco P. The effect of 18F-FDG-PET/CT respiratory gating on detected metabolic activity in lung lesions. Ann Nucl Med 2009;23:191 6. 11. Werner MK, Parker JA, Kolodny GM, English JR, Palmer MR. Respiratory gating enhances imaging of pulmonary nodules and measurement of tracer uptake in FDG PET/CT. AJR Am J Roentgenol 2009;193:1640 5. 12. García Vicente AM, Soriano Castrejón AM, Talavera Rubio MP, León Martín AA, Palomar Muñoz AM, Pilkington Woll JP, et al. (18)F-FDG PET-CT respiratory gating in characterization of pulmonary lesions: approximation towards clinical indications. Ann Nucl Med 2010;24:207 14. 13. Boellaard R, O Doherty MJ, Weber WA, Mottaghy FM, Lonsdale MN, Stroobants SG, et al. FDG PET and PET/CT: EANM procedure guidelines for tumour PET imaging: version 1.0. Eur J Nucl Med Mol Imaging 2010;37:181 200. 14. Kini VR, Vedam SS, Keall PJ, Patil S, Chen C, Mohan R. Patient training in respiratory-gated radiotherapy. Med Dosim 2003;28:7 11. 15. Otani Y, Fukuda I, Tsukamoto N, Kumazaki Y, Sekine H, Imabayashi E, et al. A comparison of the respiratory signals acquired by different respiratory monitoring systems used in respiratory gated radiotherapy. Med Phys 2010;37:6178 86. 16. Alkhawaldeh K, Bural G, Kumar R, Alavi A. Impact of dual-timepoint (18)F-FDG PET imaging and partial volume correction in the assessment of solitary pulmonary nodules. Eur J Nucl Med Mol Imaging 2008;35:246 52.