Patients
Between March 1, 2015, and July 1, 2016, 160 PRRT treatment cycles with [177Lu]-DOTA-TATE were administered to 64 patients at our institution. Inclusion criteria for this study were (a) age ≥ 20 years, (b) patients who started and completed their series of treatments in this period, and (c) patients for whom the sole reason for treatment discontinuation was treatment toxicity (hematotoxicity, although not reflected by the bone marrow dosimetry or general deterioration) or an expected absorbed dose > 25.3 Gy to the kidneys and > 2.2 Gy to the bone marrow.
Among these 64 patients, 36 of them started and completed their therapy between March 2015 and July 2016. In total, four patients were excluded because of disease progression and two were deceased before completing the series of treatment. Two patients who received a single treatment because of an insufficient or inexistent uptake in tumors were also excluded from the study. Finally, four more patients were excluded due to missing data on hospital archiving system. After all, 24 patients (12 men, 12 women; average age 61 years, range 37–85 years) were included in this single-center retrospective study (Fig. 1). The total number of therapy cycles n
trt
included in the study is 83 with a median equal to four (1 to 4 cycles).
Patient demographics are shown in Table 1.
PRRT therapy
[DOTA0,Tyr3] Octreotate was obtained either from ABX (Radeberg, Germany) or CS Bio Co. (Menlo Park, CA, USA). 177LuCl3 was supplied by PerkinElmer, Inc. (Waltham, MA, USA) and [177Lu]-DOTA-Octreotate was locally prepared by Isorad Ltd. (Soreq NRC, Yavne, Israel). Quality control for radiochemical purity was performed on each lot using high-performance liquid chromatography (HPLC) and instant thin-layer chromatography (ITLC) scanner and only labeling yields over 99% were accepted for treatment.
Infusion of amino acids (Vamin 18 g N/L electrolyte-free, Fresenius Kabi) started about 1.5 h before the administration of the radiopharmaceutical and lasted for 5 h. The radioactive ligand, diluted in 500 ml of saline, was co-administered intravenously over a period of 30 min. The mean activity per cycle of treatment was 7.4 ± 0.25 GBq (200.4 ± 6.7 mCi) with a median cumulative activity of 29.2 GBq (7.5–30.4 GBq). The interval between treatment cycles was 6–12 weeks.
Image acquisition
All 24 patients included in the study underwent a planar whole-body examination under a gamma camera after each cycle of treatment. Additionally, SPECT/CTs of the abdomen including the kidneys, liver, and spleen were acquired 18 h, 25 h, and 7 days after the injection of the first therapeutic dose in order to estimate the pharmacokinetics of [177Lu]-DOTA-TATE in these organs. For the following treatments, patients underwent a single SPECT/CT about 20 h after the administration of the radiopharmaceutical, assuming minor changes in the effective half-life for organs of interest [12]. All images were acquired on a Discovery NM/CT 670 camera with anatomical image capability (International General Electric, General Electric Medical Systems, Haifa, Israel). This system combines a dual-head coincidence SPECT camera with an axial field of view (FOV) of 40 × 54 cm, a NaI(Tl) crystal thickness of 9.5 mm, and 59 photomultiplier tubes (PMT). All functional images were acquired with a 20% energy window around the main photopeak of 177Lu (208 keV; 11% probability) [20] with medium-energy general purpose (MEGP) collimators. Whole body images were acquired with step-and-shoot mode (180 s per view; about 20 min acquisition) in a 256 × 1024 matrix, zoom 1.0, and body contour. SPECT imaging was performed applying 60 views over 360° (30 angular steps per head, 6° angle step) with a 30 s exposure per frame (15 min acquisition) in a 128 × 128 matrix size (4.4 mm pixels), zoom 1.0, and body contour. Anatomical CT images were acquired before each SPECT acquisition with the integrated BrightSpeed multidetector CT (24 rows – maximum 16 slices/rotation) using a tube voltage of 120 kV and the smart current option (80–220 mA). Calibration of SPECT images was based on a series of 29 SPECT acquisitions of a 20-mL vial placed in the center of the gamma camera FOV with a known activity of 177Lu ranging from 114.7 MBq (3.1 mCi) to 7215 MBq (195 mCi). The 177Lu calibration source was placed in the center of eight 1-L saline bags with two additional distant 20-mL 177Lu sources in order to simulate an amount of scatter similar to a clinical scan. However, no scatter correction was applied, and thus, contributions from scattered photons were ignored. No dead time was observed during calibration. This study was performed entirely independently of the camera vendor.
Blood activity concentration measurements
In order to quantify the self-dose to the bone marrow, blood samples were drawn at 18 and 25 h after the first injection of the radiopharmaceutical and after about 20 h for the following treatments. No other samples have been drawn subsequently for all the 24 patients included in this work due to organizational constraints in our department.
The samples were accurately weighed (Sartorius BL310 balance, 0.01 g precision), and the radionuclide activity concentration in the blood was measured using a NaI(Tl) well gamma-counter (Wizard 1480 3″, Perkin Elmer). The measurements were repeated three times for each blood sample. Normalization of the counter was made every 6 months, unless the instrument sends a warning for performing it. The blood activity concentration has been fitted by a mono-exponential curve and integrated to infinity in order to estimate the residence time and then the self-absorbed dose to the bone marrow, assuming that the activity concentration in the latter is the same as in the blood [18].
Image analysis and dosimetry calculation
Image analysis for dosimetry was performed using the General Electric (GE) Dosimetry toolkit (DTK) software [21] available for the Xeleris 3.0 Workstation (International General Electric, General Electric Medical Systems, Haifa, Israel). The ordered subsets expectation maximization (OSEM) algorithm with attenuation correction (from CT attenuation maps) and resolution recovery (for blurring) included in the Xeleris 3.0 workstation were used. No scatter correction was applied. In the current processing, CT is used for attenuation correction and for two types of registration: (i) registration of SPECT with CT and (ii) registration of the three SPECT/CTs (of the first cycle of treatment) one with the other. This allows the transfer of the VOIs drawn on the first SPECT/CT to the two other. DTK proposes different types of automatic or semi-automatic registrations: SPECT/CT “inherently aligned,” alignment “by center of volumes” (force the center of the SPECT raster to match the center of the CT raster), alignment “by table position” (adjust automatically according to table position during the SPECT and the CT acquisition), or alignment “by single landmark” (the user mark one spot in each modality and the rest of the volumes are registered accordingly). An additional option allows also manual adjustments of the alignment in the three dimensions (roll, pan, azimuth, and elevation) by the user. However, none of the automatic or semi-automatic registrations led to a correct registration between the first acquired CT and a SPECT acquired at a different date. Therefore, all the registrations between the first CT and the subsequent SPECTs (the first one excluded) were done manually. Using the “classical” protocol, SPECT and CT acquisitions were inherently aligned and did not need any additional adjustment. Processing includes either semi-automatic (threshold approach) or manual three-dimensional delineation of the organs on functional (SPECT) or anatomical (CT) images. VOIs were placed over the whole healthy organs of interest (kidneys, liver, spleen, and remainder of the body) and over tumors. For the kidneys and spleen, VOIs have been drawn using either the semi-automatic delineation tool on SPECT images or manual delineation on CT. The delineation method used was user- and case-dependent; in cases where the kidneys/spleen was well defined in SPECT images without surrounding tumors, the semi-automatic delineation tool has been used while in other cases, a manual delineation has been performed. It should be noted though that even when a semi-automatic method was used, the threshold of uptake for delineation was defined anew for each patient, thus increasing its user dependency. For the healthy liver and remainder of the body, VOIs have been drawn using manual delineation for all the 83 treatment cycles included in this study. For tumors, SPECT-based semi-automated segmentation has been used. Figure 2 shows an example of the drawn VOIs using the GE DTK software. It is noteworthy that the GE DTK software does not allow copying the VOIs delineated on a SPECT/CT from 1 cycle of treatment to the SPECT/CT done after another cycle of treatment. Thus, VOIs were drawn again on attenuation corrected SPECT images (using either the “classical” or the “single CT” protocol) for each cycle of treatment. Finally, the GE DTK software gives the volume of the VOIs and the activity concentrations in the organs of interest and tumors for each time point (18 h, 25 h, and 7 days post-injection for the first cycle and 20 h post-injection for the following ones). An in-house Interactive Data Language (IDL) code developed in our department is then used to obtain the absorbed radiation doses.
After the first treatment, for the kidneys, liver, spleen, remainder of the body, and tumors, residence times have been calculated as the area under the curve of a single exponential fit of the activity concentrations values given by GE DTK software at 18 h, 25 h, and 7 days post-injection. For the blood, the mono-exponential fit is based only on values at 18 and 25 h post-injection. For the following treatments, assuming an unchanged effective half-life of [177Lu]-DOTA-TATE for these organs, the activity concentrations at different times are deduced [12]. Radiation absorbed doses to the tumors were computed by using the method proposed by Sandström et al. [22] where self-doses only are taken into account. The absorbed doses (mGy) were obtained by the multiplication of the residence time of the radioactivity concentration in the tumor ([MBq ⋅ s]/[MBq ⋅ kg]) by an appropriate dose concentration factor (DCFtumor = 0.0236 [mGy ⋅ g]/[MBq ⋅ s]) and by the administered activity Aadm (MBq). Radiation absorbed doses to the kidneys, liver, spleen, bone marrow, and remainder of the body were calculated using the Medical Internal Radiation Dose (MIRD) formalism [16] where self-doses and cross-doses are considered as follows:
$$ D\left({r}_{\mathrm{t}}\right)={A}_{\mathrm{adm}}\bullet \left[{t}_{{\mathrm{r}}_{\mathrm{t}}}\bullet \mathrm{DF}\left({r}_{\mathrm{t}}\leftarrow {r}_{\mathrm{t}}\right)+\sum \limits_{\mathrm{s}\ne \mathrm{t}}{t}_{{\mathrm{r}}_{\mathrm{s}}}\bullet \mathrm{DF}\left({r}_{\mathrm{t}}\leftarrow {r}_{\mathrm{s}}\right)\right] $$
(1)
with D(rt) the dose absorbed in the target organ rt in [mGy]; \( {t}_{{\mathrm{r}}_{\mathrm{t}}} \) and \( {t}_{{\mathrm{r}}_{\mathrm{s}}} \) respectively the residence times in the target rt and source organ rs in [MBq∙s]/[MBq∙kg], and DF(b←a) the dose factor for a couple source a-target b in [mGy]/[MBq∙s]. The DFs were taken from OLINDA/EXM 1.0 [23] for the adult male and adult female phantoms. The residence times for a specific organ have been scaled for differences in mass between the drawn VOIs (considering a tissue density of 1.0 g/cm3) and the mass organ considered in OLINDA/EXM 1.0 [24]. Because most patients had a significant uptake in tumors, the latter have been taken into account for the cross-dose calculation to other organs. The DFs used for tumors are those of the liver since tumors are generally localized there. For the remainder of the body, the cross-dose contribution to the bone marrow is overestimated in OLINDA/EXM 1.0 when there is no uptake of the radionuclide in the bone, as shown previously by Stabin et al. [25]. Thus, the cross-dose DF from the remainder of the body to the bone marrow has been modified and replaced by a DF equal to 30.3 nGy∙kg/MBq∙s for the male adult phantom and 35.8 nGy∙kg/MBq∙s for the female adult phantom as given in Ref. [26].
In this work, only the absorbed doses to the kidneys and bone marrow obtained with the “classical” and “single CT” protocols are presented and compared since the dose to other organs does not influence the patient management.
Statistical methods
Absorbed dose results to the kidneys and bone marrow were analyzed in two ways. First, the Bland and Altman plots showing differences between the results obtained from the “classical” and the “single CT” protocols were built. These plots show the absolute value of the difference between results obtained with the two protocols divided by the mean of both measurements versus the mean of the measurements. Secondly, the Pearson’s correlation coefficient between data obtained from one method versus another has been computed.
Intra-observer reproducibility was determined for all the 24 patients for the kidneys and bone marrow using Bland and Altman plots and linear regression analysis to assess the intrinsic consistency of each measurement. Inter-observer reproducibility was based on the analysis of two independent series of measurements made by two examiners. Intra- and inter-observer reproducibility were estimated from measurements using the “classical” protocol in order to exclude the influence of suboptimal co-registration of SPECTs and CTs. Indeed, as said before, using the classical protocol SPECT and CT acquisitions are inherently aligned such that the examiners did not need to do any adjustment for all the 24 patients.
For all tests, P < 0.01 was considered significant.