Study design
FDG PET/computed tomography (CT) studies with a normal or near-normal activity distribution were included. All reports from clinical examinations were retrospectively evaluated in a ‘backwards’ order. Between September 2012 and October 2009, 500 examinations fulfilled our inclusion criteria. Patients who had been subjected to chemotherapy (including GSF treatment) or radiotherapy <4 weeks prior to the examination were excluded; we cannot exclude, however, that in a very few patients the available information might have been incomplete. Otherwise, patients were included irrespective of concurrent diseases and/or therapy. Examples of near-normal activity distributions were small, probably benign lung lesions or radiation portals with an FDG uptake not exceeding normal soft tissue activity, apparent reactive normal-sized lymph nodes with a moderately increased uptake, and a weak uptake in surgical scars and accumulation in minor, presumably benign skin lesions. Patients with a very small FDG extravasation at the injection site were studied, while individuals with larger extravasations were excluded. Similarly, only patients with a minute brown adipose tissue uptake were included. Patients with a visually apparent generally increased muscular uptake as previously studied by us were not included [11]. No patient with evidence of active malignancy was studied.
In the first evaluation, the correlation between the blood glucose level and SUVmean of various normal tissues was assessed in all 500 patients. In the second evaluation, patients with a B-glucose ≥7.0 mmol/l (increased) were paired with regard to age and gender with patients with a B-glucose ≤6.0 mmol/l (normal). Of 75 patients with an increased B-glucose, 62 patients could be matched with normals and were studied.
Patients
In the first evaluation, all 500 patients were studied. The mean age was 58 years (range 11 to 89). There were 248 males and 252 females. Mean B-glucose was 6.0 mmol/l (±SD 1.6). The second (paired) evaluation comprised 62 + 62 patients with a mean age of 65 years (range 25 to 80), including 33 males and 29 females in each group. In the patients with an increased B-glucose, the mean B-glucose concentration was 9.0 mmol/l (±SD 2.0) and the mean body mass index (BMI) was 26.5 kg/m2 (±SD 4.9). Of these patients, 26 were known diabetics, 13 of whom were on insulin treatment. In the controls, the mean B-glucose concentration was 5.2 mmol/l (±SD 0.6) and the mean BMI was 24.3 kg/m2 (±SD 3.3). One of these patients was a known diabetic on insulin treatment. There was a strongly significant difference of the B-glucose between the two subgroups (p < 0.0001). This retrospective investigation was approved by the regional ethics research committee (Regionala etikprövningsnämnden i Stockholm, Karolinska Institutet, SE-171 77 Stockholm, Sweden; unique identifying numbers 2009/1491-31/2 and 2012/1434-32).
PET/CT examinations
B-glucose was measured immediately prior to administration of FDG using the same glucometer, HemoCue® Glucose 201+ (Hemocue AB, Ängelholm, Sweden).
A Biograph 64 TruePoint TrueV PET/CT scanner (Siemens Medical Solutions, Erlangen, Germany) was used. The examination was performed according to the European guidelines, although patients with increased B-glucose were not rescheduled [8]. FDG (4 MBq/kg body weight) was administered intravenously (i.v.). The examination was performed approximately 1 h thereafter. It usually included the mid-skull to the proximal thigh. Prior to examination with i.v. contrast medium, a low-dose CT was performed for attenuation and scatter correction. Directly thereafter, the PET examination was done, followed by a diagnostic (full-current) CT with or without i.v. contrast medium. At examinations without i.v. contrast medium, the full-dose CT was used for attenuation and scatter correction.
The diagnostic CT examinations were performed with a tube tension of 120 kV, a pitch of 0.8, a slice thickness of 1.2 mm and a rotation speed of 0.5 s. The current was set to 160 mAs (reference), and dose modulation (CARE Dose4D) was applied. In the examinations done only for photon attenuation and scattering correction, CT was performed with a tube tension of 120 kV, a pitch of 0.8, a slice thickness of 1.2 mm, a rotation speed of 0.5 s and a current of 50 mAs. CT acquisitions were always done with the breath-holding technique at a mean inspiratory level.
There was a mean of 61 min (range 50 to 70 min) between tracer administration and PET registration. 3D PET acquisition was done for 3 min at each bed position during normal tidal breathing. The PET data were subsequently reconstructed with the manufacturer’s 2D-OSEM algorithm (four iterations and eight subsets) using a 5-mm post-reconstruction Gaussian filter. The image matrix size was 168 × 168 with a slice thickness of 5 mm. In addition to attenuation and scatter correction, all data were corrected for dead time and random coincidences.
Image analysis
Activity quantification was done in the PET images using Siemens syngo MultiModality Workplace (syngo MMWP, VE36A). Volumes of interest (VOIs) of different sizes and shapes adapted to the various organs/tissues were drawn manually, avoiding margins.
The definition in the images of the various VOIs was done by combining the information from both the CT and PET acquisitions. SUVmean of the liver was calculated by averaging the values of VOIs with a volume of 20 to 50 cm3 allocated at the centre of the right and left liver lobes, respectively. SUVmean of the spleen was similarly averaged from three elliptical VOIs with a volume of 4 to 8 cm3 in various portions of the organ. Muscular SUVmean was calculated by averaging the values of elliptical VOIs allocated in the right and left shoulder muscles (25 to 50 cm3), in both psoas muscles (5 to 10 cm3) and in the right and left gluteal muscles (25 to 50 cm3), respectively. SUVmean of the lungs was calculated by averaging the values of parasagittal, ‘flat’ VOIs with a volume of 25 to 50 cm3 in each lung, covering both the upper and lower lobes. Blood SUVmean was assessed by averaging the activity of five VOIs with a volume of 0.8 to 2 cm3 allocated in different portions of the lumen of the large mediastinal vessels. As they could not be discerned from the diffuse mediastinal FDG activity in most patients, allocation of these VOIs was done using the CT images. In cases of increased activity of the vessel wall, this was avoided. Bone marrow SUVmean was calculated by averaging the values of elliptical VOIs with a volume of 2 to 5 cm3 allocated in both iliac crests and spherical VOIs with a volume of 1 to 1.5 cm3 in each lumbar vertebral body. If a vertebra was not possible to evaluate because of a compression or extensive spondylosis, for example, it was excluded. In a few studies, one or two of the lower dorsal vertebrae were included instead.
Statistical methods
The distribution of the data showed that the correlation between B-glucose and SUVmean of the various organs/tissues could be analysed by Pearson’s correlation coefficient, applicable for linear associations. SUVmean of the muscles showed a skewed distribution (>1), which is why a reciprocal transformation was carried out prior to the analysis. Otherwise, such corrections were not done. A correlation coefficient of 0 to 0.25 indicates little or no relationship. Correlation coefficients between 0.25 and 0.50 indicate a fair degree of relationship, those between 0.50 and 0.75 indicate a moderate to good relationship and those >0.75 indicate a very good to excellent relationship [12]. The coefficient of determination (R2) indicates the proportion of the total variation in SUVmean that is explained by the variable studied, i.e. B-glucose.
The distribution of the data allowed for the use of Student’s two-tailed t test for independent samples for assessing the difference in SUVmean between the patients with increased B-glucose and the patients with normal B-glucose. The Mann–Whitney U test was applied to compare the B-glucose level of the two subgroups as the variable was not normally distributed. P < 0.05 was considered statistically significant.