The articles in this supplement were presented at the “12th Conference on Cancer Therapy with Antibodies and Immunoconjugates,” in Parsippany, New Jersey, October 16-18, 2008.
177Lu-(DOTA0,Tyr3) octreotate is a new treatment modality for disseminated neuroendocrine tumors. According to a consensus protocol, the calculated maximally tolerated absorbed dose to the kidney should not exceed 27 Gy. In commonly used dosimetry methods, planar imaging is used for determination of the residence time, whereas the kidney mass is determined from a computed tomography (CT) scan.
Three different quantification methods were used to evaluate the absorbed dose to the kidneys. The first method involved common planar activity imaging, and the absorbed dose was calculated using the medical internal radiation dose (MIRD) formalism, using CT scan-based kidney masses. For this method, 2 region of interest locations for the background correction were investigated. The second method also included single-photon emission computed tomography (SPECT) data, which were used to scale the amplitude of the time-activity curve obtained from planar images. The absorbed dose was calculated as in the planar method. The third method used quantitative SPECT images converted to absorbed dose rate images, where the median absorbed dose rate in the kidneys was calculated in a volume of interest defined over the renal cortex.
For some patients, the results showed a large difference in calculated kidney-absorbed doses, depending on the dosimetry method. The 2 SPECT-based methods generally gave consistent values, although the calculations were based on different assumptions. Dosimetry using the baseline planar method gave higher absorbed doses in all patients. The values obtained from planar imaging with a background region of interest placed adjacent to the kidneys were more consistent with dosimetry also including SPECT. For the accumulated tumor absorbed dose, the first 2 of the 4 planned therapy cycles made the major contribution.
Several options are available for the treatment of metastatic neuroendocrine tumors (NETs) including somatostatin analogs, interferon-α, cytotoxic chemotherapy, surgical resection, hepatic artery embolization, and possibly liver transplantation in certain suitable patients. The results of the treatment, particularly in patients with progressive disease, show a low objective response and short time to progression.
Peptide receptor radionuclide therapy with 177Lu-(DOTA0,Tyr3) octreotate is a new modality for treating disseminated neuroendocrine tumors with low (<2%) or moderate (2%-20%) Ki67 proliferation index. By selecting patients with somatostatin receptor-positive tumors based on 111In DTPA-octreotide scintigraphy (OctreoScan), it is possible to choose candidates suitable for subsequent 177Lu-octreotate therapy. The somatostatin analog Tyr3-octreotate differs from octreotide by the substitution of phenylalanine and threoninol with tyrosine and threonine, respectively.1 DOTA-Tyr3-octreotate has a higher affinity to the somatostatin receptor, SSTR2, than both DOTA-Tyr3-octreotide and DTPA-octreotide, as shown by Reubi et al.2177Lu emits beta-particles with a maximum energy of 149 keV, maximum particle range of 2 mm, and has a physical half-life of 6.7 days. It also emits gamma photons of 208 keV, which make it possible to perform quantitative imaging and dosimetry calculations.
The kidneys are a dose-limiting organ because of the retention of radiolabeled peptides in the renal cortex, thus leading to a highly absorbed dose, resulting in a few patients with lasting nephrotoxicity.3 The coadministration of amino acids reduces the renal uptake to some extent and, consequently, the absorbed dose to the kidneys.4
Patients treated with 177Lu-octreotate are scheduled for 3 or 4 courses, with an administered activity of 7.4 GBq per course, depending on the calculated renal-absorbed dose. According to a consensus protocol,4 the calculated maximal tolerated absorbed dose to the kidneys should not exceed 27 Gy when 177Lu-octreotate is administered. This limit is based on the “gold standard” of 23 Gy for conventionally fractionated external beam therapy, which results in a 5% probability of developing severe late kidney damage within 5 years.5 This recommendation could be questioned. Besides heterogeneities in activity uptake and absorbed dose distribution in the kidneys6 and related differences in absorbed dose rates between the 2 radiation modalities, there are different methods of calculating the absorbed dose to the kidneys and no consensus method exist.
The aims of the present study were to compare different methods of absorbed dose calculation, to determine the variation in the cumulative absorbed dose to the kidneys and to investigate how absorbed doses to tumors vary with the number of administrations. For dosimetry, the baseline method used for therapy was compared with retrospectively calculated values using 3 different dosimetry methods.
MATERIALS AND METHODS
177LuCl3 and (DOTA0,Tyr3) octreotate were obtained from IDB-Holland (Baarle Nassau, the Netherlands), and the conjugate 177Lu (DOTA0,Tyr3) octreotate was prepared at Lund University Hospital, as previously described.7
Twenty-one patients (11 female and 10 male, aged 42-81 years) suffering from disseminated NETs, with sufficiently high somatostatin receptor expression (verified by 111In-DTPA octreotide scintigraphy), were given 177Lu (DOTA0,Tyr3) octreotate. The tumor uptake of 111In-DTPA octreotide was visually judged in relation to liver uptake based on planar scintigraphy.4 The activity uptake of 111In-DTPA octreotide in the images was crucial for patient eligibility, and tumor uptake was required to at least exceed the uptake in normal liver parenchyma for this therapy to be considered. The study included patients treated since December 2006. Sixty-four treatment cycles were administered to 21 patients: 9 patients received 4 cycles, 6 patients received 3 cycles, 4 patients received 2 cycles, and 2 patients received 1 cycle each. The treatment of 2 elderly patients was concluded after 2 cycles because of intolerable hematological toxicity, and 1 patient was withdrawn because of clinically progressive disease. Of these 21 consecutively treated patients, 16 have been evaluated for dosimetry. Of these 16 patients, 8, 5, and 3 patients received 4, 3, and 2 treatment cycles, respectively.
All the 21 patients had been clinically evaluated.
All patients had previously received at least 1 other therapy regimen. The previous regimens given were biotherapy, chemotherapy, external beam radiotherapy, and/or hepatic artery embolization (see Table 1). Thus, most patients have been given several regimens before 177Lu (DOTA0,Tyr3) octreotate. The majority of the patients had progressive disease in accordance with clinical evaluation, on CT scan and 111In-octreotide scintigrams, and according to tumor markers. Eighteen patients had carcinoid tumors. Thirteen tumors were located in the midgut, 4 in the foregut, and 1 in the hindgut (rectal carcinoid). Three patients had neuroendocrine pancreas tumors. In 12 tumors, the Ki67 proliferation index was lower than 2%, and in 8 patients, the index was in the 2%-20% range (Table 1). Generally, the treatment with somatostatin analogs was interrupted 4 weeks before 177Lu administration, but in a few cases, discontinuation was contraindicated because of the manifestation of clinical symptoms. All patients gave their informed consent to participate.
Table 1. Patient Characteristics
Interferon and/or somatostatin analogs
Hepatic artery embolization
Tumor proliferation index
Ki67 2 %
Primary tumor origin
The number of treatment cycles were 3 or 4, based on a calculated maximal tolerated absorbed dose of 27 Gy to the kidneys (according to the consensus protocol). The 177Lu octreotate dosimetry method used within the therapy was based on planar imaging and kidney masses determined from CT scan (below denoted method 1B). The decision of whether a patient should receive a fourth therapy cycle was made after the third cycle, taking into account hematological and kidney toxicity and accumulative absorbed dose to the kidney. The therapy was usually stopped if the calculated absorbed dose to the kidney was predicted to exceed 27 Gy with an additional fourth cycle. In 3 cases, we considered that the benefit of a fourth therapy cycle outweighed the risk of kidney toxicity (patients numbers 6, 14, 16).
The time interval between the treatment cycles was usually 8 to 10 weeks. The maximal absorbed dose to the kidneys was 27 Gy with a few exceptions.8 Patients with slightly elevated creatinine levels (maximum 15% of upper limit) were subjected to the treatment unless their glomerular filtration rate (GFR) was <50 mL/h. The Karnofsky performance score was required to exceed 60 (ie, requiring some help, can take care of most personal requirements). After the administration of antiemetic drugs, the infusion of cationic amino acids (VAMIN-14 of 18) was started approximately 30 minutes before the co-infusion of 177Lu (DOTA0,Tyr3) octreotate via a sideline infusion pump. Routine hematology was performed before cycle 1, 4 weeks after each cycle and within 1 week before each new cycle. The liver and kidney function as well as endocrine tumor markers (p-chromogranin-A and u-5-HIAA) were monitored before each treatment cycle.
Images were acquired using a Discovery VH SPECT system (General Electric, Milwaukee, Wisconsin) including a HawkEye CT scan unit and a 1-inch NaI(Tl) crystal. For treatment cycles 1 and 2, whole body (WB) imaging was performed 0.5 hours, 24 hours, 96 hours, and 168 hours after injection. For the remaining treatment cycles, WB imaging was performed only at 24 hours and 96 hours. Complementary single photon emission computed tomography (SPECT) imaging was performed at 24 hours and/or 96 hours postinjection for all treatment cycles. Data were acquired using a 20% energy window centered over the 208 keV photopeak. The camera was equipped with a medium energy, general purpose, parallel hole collimator, and WB measurements were conducted in 384 × 1024 matrix mode with a pixel size of 2.21 mm and a scanning speed of 20 cm/min on the first imaging occasion and 10 cm/min on the following occasions. SPECT projections were acquired in 128 × 128 matrix mode with an acquisition time of 45 seconds per projection and a total of 60 projections. The pixel size was 4.02 cm, and noncircular contour-sensitive orbits were used. CT scan images were acquired using a tube voltage of 140 kV, an intensity of 2.5 mAs, and a rotation speed of 2.6 rpm. The time required was about 10 minutes, producing slices with an initial thickness of 1 cm, which was then down-sampled to 128 × 128 matrices of the same voxel size as for the SPECT data. A transmission scan showing the patient-specific variation in attenuation was made by scanning the patient using the x-ray unit of the SPECT/CT scan camera to obtain a scout image used in the attenuation correction of the planar images. The scanning time for this measurement was 2 minutes. All images were then exported to the DICOM format and processed offline using the LundADose software.9
Activity Quantification From Planar Imaging
Planar activity quantification was performed by a pixel-based method, previously described.10 The geometric mean of the measured anterior-posterior images was calculated on a pixel by pixel basis and the result was divided by the acquisition time and the system sensitivity in air. The sensitivity was determined to be 1.32 cycles per second/MBq by experimental measurements of a known amount of 177Lu activity placed in a Petri dish 10 cm from the anterior camera head. The value of the sensitivity was determined from the counts in a circular region of interest (ROI) defined around the image of the Petri dish. Attenuation correction was made by multiplying the geometric mean image to an attenuation map derived from the x-ray scout image, describing the linear attenuation coefficient for 208 keV.11 Before attenuation correction, the emission images from all imaging times were spatially registered to the attenuation map using a nonrigid, mutual information-based method, specifically developed for WB images.12 The attenuation map was also slightly smoothed to account for the different spatial resolution compared with the emission images. The measured projections were corrected for scatter before applying the geometric mean. This was performed by deconvolution10 by a scatter point spread function valid for the average thickness of the patient's abdomen, as determined from the attenuation map. The scatter point spread functions were determined by interpolation from Monte Carlo simulated scatter point spread functions, performed in elliptical water phantoms with a long axis of 30 cm and short axes ranging from 12 to 30 cm, for the current camera system, using the SIMIND Monte Carlo code.13 In the resulting activity images, obtained at each time, ROIs covering the kidneys, liver, spleen, and tumors were delineated using both the emission images and the attenuation map for visualization. The liver and spleen were delineated for the purpose of overlap correction only. Also, a narrow background ROI was defined adjacent to the kidneys. Correction for background activity and overlapping organs was then performed by the method described in Sjogreen et al,10 giving the activity in the kidneys. In the initial phase of the study, the background ROI was defined outside the abdominal cavity (denoted as method 1B below). However, it was later found that location adjacent to the kidneys (denoted as method 1A below) was more appropriate because of activity uptake in the intestines, as observed in the SPECT images.
Activity Quantification From SPECT Imaging
Transversal slices were reconstructed using an ordered subsets expectation maximization iterative reconstruction algorithm that incorporated corrections for nonhomogeneous attenuation, scattering, and collimator response.14 Eight iterations and 6 angles per subset were used. The attenuation correction was based on the measured CT scan images and calibration to mass density values using an experimentally determined bilinear function. A density threshold of 1.2 g cm−3 was used to separate soft tissue voxels from bone tissue voxels. The attenuation coefficients for soft tissue and bone tissue were 0.134 cm2 g−1 and 0.129 cm2 g−1, respectively.15 The collimator response function models geometrically collimated photons and calculate the spatial resolution degradation as a function of distance. The noncircular camera orbit was estimated from the CT scan images by calculating the maximal distance between the center of rotation and the surface of the patient plus 2 cm for each projection angle. Scatter correction was performed by means of the ESSE scatter correction method, developed by Frey et al.14 Scatter kernels for 177Lu and the current SPECT system were calculated using the SIMIND Monte Carlo program.13 The system sensitivity for the SPECT projections was experimentally determined, as described above, and found to be 8.0 cycles per second/MBq.
Calculation of Accumulated Activity and Absorbed Dose
For each treatment cycle, the ROI activity data for each imaging occasion were integrated by numerical trapezoid integration to determine the cumulated activity. From the 168 hours of examination, the kinetics was assumed to be monoexponential, as fitted to the 2 preceding data points. In the case of treatment cycles 3 and 4, where imaging was performed at only 2 points in time, the missing data points were obtained by scaling the corresponding data point from cycles 1 and 2 with the corresponding administered activity before integration.
The absorbed dose to the kidneys was evaluated using 3 different quantification methods. In methods 1A and 1B, planar activity data were used and the absorbed dose was calculated using the MIRD formalism16 using different background ROIs. In method 1A, the background ROI was defined adjacent to the kidneys, whereas in method 1B, it was defined outside the abdominal cavity. In method 2, the absorbed dose obtained in method 1A (Dmethod1A) was scaled using the activity data points from SPECT (ASPECT) and the corresponding data point from the planar images (Aplanar) as:
For SPECT images volumes of interest (VOIs) for the kidneys were constructed by slice by slice segmentation in the transversal SPECT images with guidance from the CT scan images. The VOIs were defined so as to cover the entire kidneys (including the renal pelvis and medulla). The total activity in the VOI was then determined. For both methods 1 and 2, the S value of the total body cross-dose was determined for the standard phantoms17 and linearly scaled to account for the patient's weight. For the self-dose, the S values were scaled by the individual patient kidney mass determined from the CT scan images by assuming a mass density value18 of 1.05 g/cm3.
In method 3, absorbed energy images were calculated from the SPECT activity images by considering only the energy deposition in each voxel, because the β-particle ranges are less than the dimension of the voxel. No explicit Monte Carlo simulations of electrons or photons were, thus, performed. Absorbed dose rate images were calculated by dividing each voxel value by the weight of that particular voxel (as determined from the CT scan image). A threshold of 50% of the maximum absorbed dose rate was then applied to the VOI to correct for the resolution-induced spill out of the camera system. The median absorbed dose rate was integrated by numerical trapezoid integration to give the value of the absorbed dose.
Method 1B was the one used for therapy, whereas methods 1A, 2, and 3 were evaluated retrospectively.
The 177Lu (DOTA0,Tyr3) octreotate administrations were well-tolerated without serious adverse events. Nausea occurred in 30% of the treatment cycles (National Cancer Institute Common terminology Criteria for Adverse Events, version 3, grade 2), vomiting in 10% within 24 hours after therapy (CTC grade 2), and abdominal pain in 6%. All these signs and symptoms were transitory. Hematological toxicity was mild with CTC grade 3 neutrocytopenia and thrombocytopenia in 2 elderly patients only (Table 2). Thus, hematologic toxicity did not seem to be restrictive for the therapy.
Objective response after at least 10 months was evaluated in 12 patients according to the RECIST criteria. Partial response was observed in 2 patients, minor response (25% to 50% of tumor reduction) in 3 patients, and stable disease in 5 patients with previously progressive disease. In 2 patients the radiological signs of progressive disease occurred 6 months to 8 months after the last administration. The performance was markedly improved in 10 patients (85%).
Tumor markers (Cg-A and 5-HIAA) increased in 85% of the treated patients after the first and the second therapy cycles, and then gradually decreased after administration of the fourth cycle. The level of normalization of the tumor markers was correlated to the objective response of the tumor.
The values of the absorbed dose calculated with the different methods are shown for the 16 evaluated patients in Figure 1. The median absorbed dose to the kidneys was 26 Gy (range,17-45) according to the planar method 1B used for therapy planing. The corresponding values based on, eg, method 3, were 21 Gy (range, 14-32 Gy). The average (±1 SD) absorbed dose to the kidneys per unit administered activity were 0.97 (0.24), 1.15 (0.29), 0.81 (0.21), and 0.90 (0.21) mGy/MBq for methods 1A, 1B, 2, and 3, respectively. The major cause for the difference between the planar-based methods (1A and 1B) and those including SPECT data (2 and 3) was the overlap of local high activity areas in the image area of the intestines, which gave an overestimation of the kidney activity. An example of this can be seen in Figure 2, where an activity of 5.6 MBq was measured within the kidney from the SPECT image, whereas 10.2 MBq was measured in a planar image of the same projected ROI area. By placing the background region adjacent to the kidney (method 1A), the values from method 1B were reduced on the average by 18%.
In 12 evaluated patients who have completed the treatments (3 or 4 cycles), the creatinine levels were unchanged after a follow-up time ranging between 4 and 24 months. Seven of these patients received an absorbed dose exceeding 27 Gy to the kidneys according to dosimetry method 1B, and 4 of these (Patients 6, 11, 14, and 16) received more than 30 Gy, ie, 45, 35, 33, and 40 Gy, respectively (Fig. 1). The corresponding values according to method 3 for these patients were 32, 18, 19, and 31 Gy.
Figure 3 shows total absorbed dose per unit administered activity, using data from all therapy cycles compared with that obtained in the first therapy cycle only. The different cycles contributed on average equally to the absorbed dose (within 10%). In 2 cases, however, the contribution to the accumulated absorbed dose from 1 treatment cycle to another varied by 20%, mainly because of differences in the kinetics between the cycles.
The average total body-absorbed dose was 1.9 ± 0.7 Gy. Normalized to the administered activity, it was found to be 0.07 ± 0.02 mGy/MBq.
The absorbed doses to the tumor were calculated using method 3 for 7 of the patients. The median absorbed dose per unit administered activity was 6.7 mGy/MBq (range, 0.1-20 mGy/MBq), and the median accumulated absorbed dose was 207 Gy (range, 17-387 Gy). The contribution from the different therapy cycles to the accumulated absorbed dose to the tumor is shown in Figure 4. The first 2 of the 4 planned therapy cycles made the major contribution to the tumor absorbed dose.
According to a consensus protocol, the calculated maximally tolerated absorbed dose to the kidneys is 27 Gy when 177Lu-octreotate is administered as 3 or 4 fractions at intervals of 8-10 weeks.4 This absorbed dose limit is based on the gold standard for conventionally fractionated external beam therapy of 23 Gy, which results in a 5% probability of developing severe late kidney damage within 5 years.5 We consider that the biological effects cannot be directly applied to low dose-rate radiation, especially regarding tissues with low alpha/beta values, such as the kidneys.19 The method used for kidney dosimetry, when reported in the literature, is commonly based on planar imaging. Bodei et al20 have shown that patients with risk factors, such as hypertension and diabetes, should undergo thorough dosimetric investigation before peptide therapy with 90Y-DOTATOC and not receive a biological effective dose (BED) higher than 28 Gy, while patients with no risk factors might have a renal BED threshold of 40 Gy. Data regarding 177Lu-(DOTA0,Tyr3) octreotate dosimetry and long-term kidney toxicity are still very sparse, but the level of kidney toxicity might be at least as high as that for 90Y-DOTATOC.
In the present study, we have focused on kidney dosimetry by evaluating the variation in the calculated absorbed dose to the kidneys obtained from 3 different methods (Fig. 1). In addition, we observed that the major part of the absorbed dose to the tumor was achieved during the first 2 cycles (Fig. 3). We have also described data regarding toxicity, patient characteristics, including previous regimens, to get information of the tolerability of the 177Lu- octreotate therapy given. Taking into consideration that these patients have received several prior regimens (Table 1), there is an objective tumor response of this treatment, but there is a need of improving the efficacy.
The results of the planar activity quantification were very dependent on the location of the ROI for background correction (Fig. 2). The location of the ROI in the abdomen close to the kidneys (method 1A) partially compensated for contribution from overlapping activity in the intestines. When placing the background ROI in an area outside the abdominal cavity with a more homogeneous activity uptake (method 1B), the resulting kidney activity values were on average 18% higher. Method 1A was established retrospectively through comparing with activity distributions obtained with SPECT. Despite corrections for activity background and overlap, the absorbed dose to the kidneys as calculated by method 1A was, on average, 20% higher than that obtained with method 2. The reason for this deviation was the movement of regional high activity hotspots in the intestine during the treatment cycle, which was difficult to detect in the planar images.
Method 3 represents an independent dosimetry method, as it is not based on S values and, therefore, requires no measurement of the kidney mass. The size and shape of the VOI were, however, found to be important parameters. In this method, the VOI was defined using a 50% lower threshold value to avoid including voxels that were affected by partial volume effects because of the limited spatial resolution of the camera system. More elaborate ways may be feasible, eg, by applying a dedicated correction method for the partial volume effect. Furthermore, because of the partial volume effect, method 3 was believed to underestimate the absorbed dose to small tumors, and a calculation based on the maximum absorbed dose rate instead of the median absorbed dose rate may provide a more accurate calculation of the absorbed dose.
The present study has shown that large differences can be obtained in the calculated value of the absorbed dose to the kidneys depending on the method used for the absorbed dose calculation. The results in the present study, thus, stress the importance of describing the method used to calculate the absorbed dose.
According to Figure 4, the contribution to the accumulative absorbed dose to the tumor after each therapy cycle was gradually reduced. Because neuroendocrine tumors are generally considered to be relatively radiation-resistant, and because our results show that the first 2 cycles make the major contribution to the absorbed dose to the tumor, it appears to be appropriate to base the treatment on 2 cycles with higher administered activity, instead of 4 cycles, for patients with more rapidly growing metastases. By administering 2 therapy cycles with an increased activity within an interval of 8-10 weeks, there might be a chance to better overcome repair of sublethal cell damage and repopulation. A rapid shrinkage of these relatively radioresistant tumors is a less likely explanation to the gradually reduced uptake. We assume saturation or down-regulation of the peptide receptors is a more plausible explanation. Hematologic toxicity, kidney toxicity, and other experienced toxicity in present study did not appear to indicate any restriction for a more aggressive approach. This strategy would be especially useful in patients not suitable for cytostatic regimens and those with more aggressive tumors where there is not the time for more protracted therapy, but then taking into account the increased risk of late kidney toxicity. We, therefore, raise the question of whether the recent standard protocol for 177Lu octreotate should always be used, especially in cases of progressive metastatic diseases, bearing in mind other existing, feasible modalities.
CONFLICT OF INTEREST DISCLOSURES
The articles in this supplement represent proceedings of the “12th Conference on Cancer Therapy with Antibodies and Immunoconjugates,” in Parsippany, New Jersey, October 16-18, 2008. Unrestricted grant support for the conference was provided by Actinium Pharmaceuticals, Bayer Schering Pharma, Center for Molecular Medicine and Immunology, ImClone Systems Corporation, MDS Nordion, National Cancer Institute, National Institutes of Health, New Jersey Commission on Cancer Research, and PerkinElmer Life & Analytical Sciences. The supplement was supported by an unrestricted educational grant from ImClone Systems Corporation, a wholly owned subsidiary of Eli Lilly and Company, and by page charges to the authors. Supported by grants from The Swedish Cancer Society; Mrs. Berta Kamprad's Foundation; Gunnar Nilsson's Foundation; The Lund University Hospital Funds; and The Swedish Research Council.