Blood flow measurements
To determine the blood flow within the glandular tissue, an established methodology was applied  on three healthy test persons, two males of ages 38 and 47 and one female of age 23. Duplex sonography was conducted on these three subjects using an iU22 xMATRIX ultrasound system (Philips Healthcare, Hamburg, Germany). Initially, the salivary glands were imaged without stimulation and thereafter stimulated by either chewing gum base or sucking on lemon juice. The submandibular gland's sonograms were used representatively because they allowed a better visualization with ultrasound than the parotid glands . Duplex ultrasonography was performed during sucking on a slice of lemon. In contrast, since imaging was difficult to perform during lower jaw movements, the images were acquired shortly after a period of 1 min of chewing-gum stimulation. For the comparison of the different stimulation types, the sonograms of the non-stimulated glands served as the reference standard.
Our department has routinely been conducting pre-therapy 124I PET/CT scans for high-risk patients with differentiated thyroid cancer to perform lesion and organ-at-risk dosimetries . Ten thyroidectomized patients prior to their first radioiodine therapy with histologically confirmed advanced differentiated thyroid cancer were included. Patients were not included if they had a medical history of salivary gland disease or affection, had received external beam radiotherapy to the neck or head, were taking anticholinergic medication, or had dental implants. Furthermore, patients bearing iodine-avid lymph node metastases near the salivary glands or thyroid remnant tissue with high uptake were not included, to avoid impairments in the absorbed dose calculations . The standard patient preparation before 124I administration has been described previously . Thyroid-stimulating hormone (TSH) stimulation was achieved by withdrawal of thyroid hormone. TSH levels before imaging were ≥25 mIU/mL. The mean ± standard deviation (SD) of the 124I activity was 22.7 ± 1.0 MBq. The patients gave written informed consent to perform the examination and the study was conducted in full accordance with regional ethical committee standard.
Chewing-gum stimulation protocol
During the first day, the patients were instructed to chew on tasteless gum base (Solsona; Cafosa Gum S.A., Barcelona, Spain) approximately 20 min after capsule ingestion and to drink water. The patients were instructed to continue gum chewing with a frequency of 3 to 4 gums per hour and to drink at least 2 L of water. The patients ate lunch after the first 4-h 124I PET/CT investigation and later ate a snack and dinner. The intake of sour foods or drinks such as orange juice was avoided. The patients documented the time points of gum and water intakes during the first day.
Salivary gland dosimetry protocol
The patients underwent a series of three 124I PET/CT scans on a Biograph Duo system (Siemens Medical Solutions, Chicago, IL, USA). PET imaging was acquired at about 4, 24, and ≥96 h after oral intake of a capsule containing [124I]NaI. The emission time was 300 s per bed position. The low-dose CT acquisition parameters were as follows: tube current time product of 15 mAs, tube voltage of 110 kVp, a pitch of 1.6, and a slice thickness of 5 mm. The emission images were reconstructed using the iterative attenuation-weighted ordered-subset expectation maximization algorithm with 4 iterations and 16 subsets. A post-reconstruction three-dimensional Gaussian smoothing filter of 5 mm was applied. Standard scatter, attenuation, and dead-time corrections provided by the manufacturer were used. The reconstructed transverse emission images had a voxel size of 1.7 × 1.7 × 2.4 mm3. The measured reconstructed PET spatial resolution (expressed as full width at half maximum) was 8.2 mm. The CT images were reconstructed using a reconstruction interval of 2.4 mm; the reconstructed image had a voxel size of 1.0 × 1.0 × 2.4 mm3.
The details of the patient analyses have been described in the literature . The analyses consisted of image co-registration, delineation of the salivary gland volume, applying volume-dependent isovolume 124I recovery coefficients, and determination of the glandular residence times and absorbed doses per administered activity. Image co-registration across time was performed by matching the reference 4-h CT image with the 24-h CT image and the last CT image at ≥96 h. The resulting transformation parameters were then applied to the respective PET image for each time point. The salivary gland volumes were determined from the anatomical images by drawing a volume of interest around each salivary gland. The individual volume of interest was projected onto the co-registered PET image and used to determine the ratio of total activity to the gland volume; this ratio is referred to as the (imaged) isovolume activity concentration. The isovolume activity concentrations were then corrected for partial volume effect using isovolume 124I recovery coefficients derived from measurements with a phantom containing spheres of different volumes. It has been shown that the isovolume 124I recovery coefficients of spheres matched well with spheroids mimicking salivary glands. Finally, the radioactive decay law was applied to project the 124I activity concentration to the therapy nuclide. The 124I and projected 131I uptake values, U124(t) and U131(t) at the time point t, respectively, were determined.
The absorbed doses per 131I activity were estimated using a three-point model consisting of two contributions. The uptake curve was parameterized using a combination of linear functions from 0 to 24 h (first contribution) and a mono-exponential function from 24 h to infinity (second contribution). In the first part, trapezoid integration was applied for the area under the curve from zero time, U131(0) = 0, to 24 h, U131(24 h). The contribution resulting from the exponential phase was the time integral from 24 h to infinity using an effective 131I half-life that was calculated from the uptake values U131(24 h) and U131(≥96 h) by linear regression analysis. In the estimation of the glandular absorbed dose per activity, the volume, the residence time (see above), and the glandular tissue density were required. A density of 1.05 g/mL was used for the salivary glands. A uniform activity distribution within the salivary gland was assumed.
In the previous studies ,, the estimation of the absorbed doses per activity for the non-stimulation and lemon-juice stimulation groups was calculated using a seven-point model. In particular, additional PET measurements were acquired at 0.5, 1, 2, and 48 h after 124I administration, which were taken into account in the estimation of the residence time. The area under the uptake curve was determined by applying the trapezoid integration rule with a truncation at the time point of the last PET scan (≥96 h). To examine the effect of the reduced number of time points (three points vs. seven points) and the different integration approaches (piecewise trapezoid-exponential integration vs. trapezoid integration), the glandular absorbed doses per activity for the non-stimulation and lemon-juice stimulation groups were re-assessed using the three-point model and compared to those of the seven-point model.
The descriptive statistics included the mean, the median, the SD, the minimum, and the maximum, which are provided in the following form: mean (median) ± SD (minimum to maximum). Differences among the groups were evaluated by the Mann-Whitney U test (non-parametric test). The correlation between the different quantities was evaluated using Spearman's rank correlation coefficient. A significance level (P value) of less than 5% was considered statistically significant.