2.1 Initial phantom study
Activity estimation in the extravasation area of the patient images was realized using a local quantification protocol initially developed for renal dosimetry with 177Lu. Two phantom studies had been performed prior to the incident : one to set a calibration factor F in counts/MBq/s for 177Lu and one to evaluate its accuracy on SPECT/CT images.
The calibration phantom is a body shape phantom (PTW Body phantom ‘B’) with a cover containing 3 cylindrical inserts (PTW cover D with 3 cylinders), of which 2 are fillable and one is made of Polytetrafluoroethylene to simulate bone. The inner diameter of the cylinders was 4.6 cm. Volumes of the fillable cylinders were determined by weighing and were respectively 305 and 321 mL. The volume of the inserts were slightly larger, yet in the range of magnitude of normal kidney volume, 202 mL for men and 154 mL for women (7). The two cylinder inserts (Figure 1) were filled with 177Lu-DOTATATE and placed in the body shape phantom filled with water without activity.
The verification phantoms used (Figure 2) were a cylindrical phantom (PTW Head Phantom ‘H’) and the body shape phantom previously described, alternatively changed from one acquisition to the other. A heart insert made of two non-axial cylinders was used (PTW Heart phantom ‘C’). Only the inner cylinder, with an inner diameter of 4.4 cm and a length of 10 cm (weighted volume 150 mL) was filled with 177Lu-DOTATATE. The outer cylinder was filled with water and placed in the cylinder or body phantom filled with water.
A reference activity concentration (AC) of 1 MBq/mL was considered, corresponding to a value measured in a patient kidney 24h after 177Lu-DOTATATE, as reported in the literature (8). Six acquisitions around 7 days apart were performed for each phantom, to cover a wide range of activity concentration (AC) around this reference value. All images were acquired on a Symbia T2® camera (Siemens Healthcare) with a 5/8’’ crystal and a MELP collimator. Images were acquired and reconstructed according to the parameters detailed in Table 2 and 3, which were determined using joint EANM/MIRD recommendations for quantitative 177Lu SPECT (3). Images were corrected for attenuation using CT and for scatter using a double energy window method.
Activities were determined using a dose calibrator, calibrated for 177Lu (MEDI 405®, Veenstra). The total activities in the calibration and verification inserts and the corresponding activity concentrations are reported in Table 1.
Acquisition
|
Calibration – Cylinder 1
|
Calibration – Cylinder 2
|
Verification – Cylinder
|
A (MBq)
|
AC (MBq/mL)
|
A (MBq)
|
AC (MBq/mL)
|
A (MBq)
|
AC (MBq/mL)
|
1
|
2275
|
7.51
|
1136
|
3.54
|
471
|
3.14
|
2
|
1700
|
5.61
|
849
|
2.65
|
354
|
2.36
|
3
|
808
|
2.67
|
403
|
1.26
|
167
|
1.12
|
4
|
388
|
1.28
|
194
|
0.60
|
80
|
0.54
|
5
|
186
|
0.61
|
93
|
0.29
|
39
|
0.26
|
6
|
90
|
0.30
|
45
|
0.14
|
19
|
0.12
|
Table 1. Activities and activity concentrations in the cylinders used for calibration and verification
|
SPECT acquisition
|
Spectrometric window
|
Main: 208 keV +/- 7.5%
Scatter: 180 keV +/- 6%
|
Acquisition conditions
|
128 x 128 matrix
2 x 30 projections of 30s
Auto-contour
|
Table 2. Acquisition parameters for SPECT acquisitions
Reconstruction Algorithm
|
Flash 3D ®
|
Iterations
|
10
|
Subsets
|
10
|
Post-filtering
|
Gaussian filter with 4 mm FWHM
|
Table 3. Reconstruction parameters for SPECT images
Images were processed using Matlab® (Mathworks). The number of counts was calculated in a volume of interest (VOI) equal to the known volume of the cylinders on the calibration images. The number of counts was divided by the acquisition duration and plotted against the activity. The calibration factor F in counts/s/MBq was determined using a linear regression on this plot.
In the verification step, we tried to simulate a patient image processing: VOI was manually determined slice by slice on the CT images of the phantom and reported on SPECT images. The number of counts in the VOI was calculated and converted into activity using F. The calculated activity was compared to the known activity.
2.2 Patient
Extravasation occurred during the first cycle of treatment in a 70 year-old male patient, previously operated for small intestine neuro endocrine tumors, with several metastatic lesions in the liver, bone lesions and one subclavicular lymph node, all expressing somatostatin receptors on 111In-pentreotide scintigraphy. Nephroprotective amino acids solution was perfused intravenously in the right arm from 10:40 without any incident. Lutathera was injected in the left median cubital vein at 11:40 (T0). The routine injection protocol was used, using a pump, pushing the saline solution in the vial. The 177Lu-DOTATATE was then pushed in a second manifold to the vein of the patient, by the pressure increase in the vial. The flow was set to 100 mL/h, and then increased to 200 and 300 mL/h. At the end of the perfusion, the nurse noticed a swelling just over the left elbow. The patient did not complain of any pain. The nurse immediately stopped the perfusion and informed the nuclear physician, who confirmed the extravasation. Medical physicist and local referee for nuclear and radiological incident were informed. The residual activity of Lutathera in the vial was measured using a dose calibrator, calibrated for 177Lu (MEDI 405®, Veenstra).
Osmogel® dressings were placed on the injection site. The patient underwent several whole body and SPECT/CT images. Adapted interventions were used to stimulate the lymphatic elimination during several hours after the detection of the extravasation, including in the night: warming and elevation of the left arm and repeated self-massages.
2.3 Imaging
Whole body images were acquired with a spectrometric window of 208 keV +/-10% and a 20 cm/min speed. The first whole body image was acquired 2 hours (T2h) after the beginning of the perfusion and 1 hour after the extravasation was detected. Following routine acquisition protocol, the patient urinated before this first acquisition.
A second whole body image was acquired 5 hours (T5h) after the beginning of the perfusion. This WB images were followed by a SPECT/CT acquisition on the arm. All SPECT/CT images were acquired with the parameters provided in Table 2 and reconstructed using the parameters specified in Table 3. Scatter and CT-based attenuation corrections were applied.
The first two whole body acquisitions and the SPECT/CT images were used to make a preliminary estimation of the dose to the arm, a necessary step for deciding whether surgical rinse might be performed to avoid irreversible deterministic effects to the patient, such as tissue necrosis.
The next morning, a WB acquisition was performed at 7:40 (T20h), followed by a SPECT/CT acquisition. A last imaging session was realized at 14:04 (T26h) with WB and SPECT/CT images to refine the first dose estimation made on the day of the incident. Imaging times and types are synthesized on Figure 3.
2.4 Volume estimation
After extravasation, the 177Lu-DOTATATE spread heterogeneously in the subcutaneous tissue. Due to very low contrast, CT could not be used to determine the extravasation volume. For the initial dose estimation, the extravasation volume was determined using a 3D threshold on a ®Syngo Via software (Siemens).
For the second dose estimation, we tried to get a better idea of the dose distribution by defining 3 volumes of interest using 3D threshold on SPECT images: one large volume encompassing the extravasation area and including low uptake voxels, one medium volume close to the volume used for the initial dose estimation and one smaller volume corresponding to the voxels with higher uptake. This approach was used to get a range of dose estimations, more informative than one single value.
2.5 Dose calculation
The MIRD formalism was used to determine the dose in the extravasated area (9). Given the low energy of beta emissions from 177Lu and the high uptake of the extravasation region, only self-irradiation was considered for dose calculation.
For the first dose estimation, the percentage of activity in the arm related to the activity in the whole body was used to estimate the activity in the extravasated area. This calculation was made on the geometrical mean on anterior and posterior WB T2h images. WB images were processed with ImageJ v1.51n (10) and SPECT/CT images were processed on Syngo® software (Siemens).
T5h SPECT/CT images were used to determine the volume of the extravasated area. Effective period was determined on ROI drawn around the arm on the geometrical mean of T2h and T5h WB images.
For the second more accurate dose estimation taking all images into account, activity in the extravasated area was determined using the calibration factor estimated on phantoms data for SPECT images. VOI were defined on SPECT images.
As no SPECT/CT images were acquired at T2h. To get an estimation of the activity at this time point, ROI around the extravasation were drawn on geometric mean of whole body images acquired at T2h and T5h. A ratio between counts in ROI at T2h and T5h was calculated and applied to activity estimated on SPECT CT at T5h to approximate the activity at T2h, as shown in equation 1.
Effective period was determined with activity estimated on SPECT images and ASPECT T2h*.
For both dose estimations, cumulated activity was determined as the area under the curve of the mono exponential fit of activity calculated on images.
For dose calculation, we used the approach detailed by Sandström et al for kidney 177Lu-DOTATATE dosimetry based on a dose factor (DF), which is the absorbed energy per time-integrated activity concentration in nGy.kg/(MBq.s) (8). DF factors calculation are based on Radar website data (11). DF values are between 23.9 and 24.8 nGy.kg/(MBq.s) for spheres between 100g and 2000g. We chose to use a unique approximate value of 24 nGy.kg/(MBq.s) for dose estimations.